<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><htmlxmlns="http://www.w3.org/1999/xhtml"><head><metahttp-equiv="Content-Type"content="text/html; charset=UTF-8"/><title>TensorFlow.Ops</title><linkhref="ocean.css"rel="stylesheet"type="text/css"title="Ocean"/><scriptsrc="haddock-util.js"type="text/javascript"></script><scripttype="text/javascript">//<![CDATA[
window.onload = function () {pageLoad();setSynopsis("mini_TensorFlow-Ops.html");};
</script></head><body><divid="package-header"><ulclass="links"id="page-menu"><li><ahref="src/TensorFlow-Ops.html">Source</a></li><li><ahref="index.html">Contents</a></li><li><ahref="doc-index.html">Index</a></li></ul><pclass="caption">tensorflow-ops-0.1.0.0: Friendly layer around TensorFlow bindings.</p></div><divid="content"><divid="module-header"><tableclass="info"><tr><th>Safe Haskell</th><td>None</td></tr><tr><th>Language</th><td>Haskell2010</td></tr></table><pclass="caption">TensorFlow.Ops</p></div><divid="description"><pclass="caption">Description</p><divclass="doc"><p>This module contains definitions for some built-in TensorFlow operations.</p><p>Note that certain, "stateful" ops like <code>variable</code> and <code>assign</code> return a
<code><ahref="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a></code> action (e.g., <code>Build (Tensor Ref a)</code> instead of a pure value; the
returned <code><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code>s are always rendered in the current <code><ahref="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a></code> context. This
approach helps us avoid problems with inlining or common subexpression
elimination, by writing</p><pre>do
v <- variable []
w <- assign v 3
render $ w * w</pre><p>instead of</p><pre>let
v = variable []
w = assign v 3
in w * w</pre><p>since the latter could be reasonably transformed by the compiler into (or
vice versa)</p><pre>let
v = variable []
w = assign v 3
w' = assign v 3
in w * w'</pre><p>Ops should return a <code><ahref="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a></code> action if their original <code>OpDef</code> marks them as
stateful, or if they take any Refs as input. (This mirrors the rules that
of the input Tensor to reduce across. For vectors, use dimension = 0.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a><ahref="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><tdclass="doc"><p><strong>output</strong></p></td></tr></table></div><divclass="doc"><p>Returns the index with the largest value across dimensions of a tensor.</p></div></div><divclass="top"><pclass="src"><aname="v:assign"class="def">assign</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> t</td><tdclass="doc"><p><strong>ref</strong>: Should be from a <code>Variable</code> node. May be uninitialized.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><tdclass="doc"><p><strong>value</strong>: The value to be assigned to the variable.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> t)</td><tdclass="doc"><p><strong>output_ref</strong>: = Same as "ref". Returned as a convenience for operations that want
to use the new value after the variable has been reset.</p></td></tr></table></div><divclass="doc"><p>Update <code>ref</code> by assigning <code><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code> to it.</p><p>This operation outputs "ref" after the assignment is done.
This makes it easier to chain operations that need to use the reset value.</p></div></div><divclass="top"><pclass="src"><aname="v:broadcastGradientArgs"class="def">broadcastGradientArgs</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: (<ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) t)</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><tdclass="doc"><p><strong>s0</strong></p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><tdclass="doc"><p><strong>s1</strong></p></td></tr><tr><tdclass="src">-> (<ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</td><tdclass="doc"><p>(<strong>r0</strong>, <strong>r1</strong>)</p><ul><li><strong>r0</strong></li><li><strong>r1</strong></li></ul></td></tr></table></div><divclass="doc"><p>Return the reduction indices for computing gradients of s0 op s1 with broadcast.</p><p>This is typically used by gradient computations for a broadcasting operation.</p></div></div><divclass="top"><pclass="src"><aname="v:cast"class="def">cast</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: (<ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> srcT, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dstT)</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 srcT</td><tdclass="doc"><p><strong>x</strong></p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> dstT</td><tdclass="doc"><p><strong>y</strong></p></td></tr></table></div><divclass="doc"><p>Cast x of type SrcT to y of DstT.</p></div></div><divclass="top"><pclass="src"><aname="v:concat"class="def">concat</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <ahref="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><tdclass="doc"><p><strong>concat_dim</strong>: 0-D. The dimension along which to concatenate. Must be in the
range [0, rank(values)).</p></td></tr><tr><tdclass="src">-> [<ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t]</td><tdclass="doc"><p><strong>values</strong>: The <code>N</code> Tensors to concatenate. Their ranks and types must match,
and their sizes must match in all dimensions except <code>concat_dim</code>.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><tdclass="doc"><p><strong>output</strong>: A <code><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a></code> with the concatenation of values stacked along the
<code>concat_dim</code> dimension. This tensor's shape matches that of <code>values</code> except
in <code>concat_dim</code> where it has the sum of the sizes.</p></td></tr></table></div><divclass="doc"><p>Concatenates tensors along one dimension.</p></div></div><divclass="top"><pclass="src"><aname="v:constant"class="def">constant</a> :: <spanclass="keyword">forall</span> a. <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =><ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:Shape">Shape</a> -> [a] -><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a <ahref="src/TensorFlow-Ops.html#constant"class="link">Source</a></p><divclass="doc"><p>Create a constant tensor.</p><p>The values should be in row major order, e.g.,</p><p>element 0: index (0, ..., 0)
<ahref="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><divclass="top"><pclass="src"><aname="v:expandDims"class="def">expandDims</a> :: <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <ahref="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> -><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t <ahref="src/TensorFlow-Ops.html#expandDims"class="link">Source</a></p></div><divclass="top"><pclass="src"><aname="v:initializedVariable"class="def">initializedVariable</a> :: <spanclass="keyword">forall</span> a. <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a -><ahref="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a) <ahref="src/TensorFlow-Ops.html#initializedVariable"class="link">Source</a></p><divclass="doc"><p>Creates a variable initialized to the given value.
Initialization happens next time session runs.</p></div></div><divclass="top"><pclass="src"><aname="v:zeroInitializedVariable"class="def">zeroInitializedVariable</a> :: (<ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a, <ahref="../base-4.8.2.0/Prelude.html#t:Num">Num</a> a) =><ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:Shape">Shape</a> -><ahref="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a) <ahref="src/TensorFlow-Ops.html#zeroInitializedVariable"class="link">Source</a></p><divclass="doc"><p>Creates a zero-initialized variable with the given shape.</p></div></div><divclass="top"><pclass="src"><aname="v:fill"class="def">fill</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 <ahref="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><tdclass="doc"><p><strong>dims</strong>: 1-D. Represents the shape of the output tensor.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><tdclass="doc"><p><strong>value</strong>: 0-D (scalar). Value to fill the returned tensor.</p><p><code>compatibility(numpy)
Equivalent to np.full
</code>end_compatibility</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><tdclass="doc"><p><strong>output</strong></p></td></tr></table></div><divclass="doc"><p>Creates a tensor filled with a scalar value.</p><p>This operation creates a tensor of shape <code>dims</code> and fills it with <code><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#v:value">value</a></code>.</p><p>For example:</p><p>```prettyprint
```</p></div></div><divclass="top"><pclass="src"><aname="v:oneHot"class="def">oneHot</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: (<ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tI, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <ahref="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> ([] *)))) tI)</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tI</td><tdclass="doc"><p><strong>indices</strong>: A tensor of indices.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 <ahref="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a></td><tdclass="doc"><p><strong>depth</strong>: A scalar defining the depth of the one hot dimension.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><tdclass="doc"><p><strong>on_value</strong>: A scalar defining the value to fill in output when `indices[j] = i`.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><tdclass="doc"><p><strong>off_value</strong>: A scalar defining the value to fill in output when `indices[j] != i`.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><tdclass="doc"><p><strong>output</strong>: The one-hot tensor.</p></td></tr></table></div><divclass="doc"><p>Returns a one-hot tensor.</p><p>The locations represented by indices in <code>indices</code> take value <code>on_value</code>,
while all other locations take value <code>off_value</code>.</p><p>If the input <code>indices</code> is rank <code>N</code>, the output will have rank `N+1`,
The new axis is created at dimension <code>axis</code> (default: the new axis is
appended at the end).</p><p>If <code>indices</code> is a scalar the output shape will be a vector of length <code>depth</code>.</p><p>If <code>indices</code> is a vector of length <code>features</code>, the output shape will be:
```
features x depth if axis == -1
depth x features if axis == 0
```</p><p>If <code>indices</code> is a matrix (batch) with shape `[batch, features]`,
the output shape will be:
```
batch x features x depth if axis == -1
batch x depth x features if axis == 1
depth x batch x features if axis == 0
```</p><p>Examples
=========</p><p>Suppose that</p><p>```
indices = [0, 2, -1, 1]
depth = 3
on_value = 5.0
off_value = 0.0
axis = -1
```</p><p>Then output is `[4 x 3]`:</p><p>```output =
[5.0 0.0 0.0] // one_hot(0)
[0.0 0.0 5.0] // one_hot(2)
[0.0 0.0 0.0] // one_hot(-1)
[0.0 5.0 0.0] // one_hot(1)
```</p><p>Suppose that</p><p>```
indices = [0, 2, -1, 1]
depth = 3
on_value = 0.0
off_value = 3.0
axis = 0
```</p><p>Then output is `[3 x 4]`:</p><p>```output =
[0.0 3.0 3.0 3.0]
[3.0 3.0 3.0 0.0]
[3.0 3.0 3.0 3.0]
[3.0 0.0 3.0 3.0]
// ^ one_hot(0)
// ^ one_hot(2)
// ^ one_hot(-1)
// ^ one_hot(1)
```
Suppose that</p><p>```
indices = [[0, 2], [1, -1]]
depth = 3
on_value = 1.0
off_value = 0.0
axis = -1
```</p><p>Then output is `[2 x 2 x 3]`:</p><p>```output =
[
[1.0, 0.0, 0.0] // one_hot(0)
[0.0, 0.0, 1.0] // one_hot(2)
][
[0.0, 1.0, 0.0] // one_hot(1)
[0.0, 0.0, 0.0] // one_hot(-1)
]```</p></div></div><divclass="top"><pclass="src"><aname="v:matMul"class="def">matMul</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: (<ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<ahref="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a><ahref="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<ahref="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a><ahref="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <ahref="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <ahref="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <ahref="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))))) t)</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><tdclass="doc"><p><strong>a</strong></p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><tdclass="doc"><p><strong>b</strong></p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><tdclass="doc"><p><strong>product</strong></p></td></tr></table></div><divclass="doc"><p>Multiply the matrix "a" by the matrix "b".</p><p>The inputs must be two-dimensional matrices and the inner dimension of
cublas.</li></ul></div></div><divclass="top"><pclass="src"><aname="v:matTranspose"class="def">matTranspose</a> :: <spanclass="keyword">forall</span> a v. <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a <ahref="src/TensorFlow-Ops.html#matTranspose"class="link">Source</a></p></div><divclass="top"><pclass="src"><aname="v:mean"class="def">mean</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: (<ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<ahref="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a><ahref="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<ahref="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a><ahref="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int16">Int16</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int8">Int8</a> ((:) * <ahref="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <ahref="../base-4.8.2.0/Data-Word.html#t:Word8">Word8</a> ((:) * <ahref="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <ahref="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))))))))) t, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) tidx)</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><tdclass="doc"><p><strong>input</strong>: The tensor to reduce.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><tdclass="doc"><p><strong>reduction_indices</strong>: The dimensions to reduce.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><tdclass="doc"><p><strong>output</strong>: The reduced tensor.</p></td></tr></table></div><divclass="doc"><p>Computes the mean of elements across dimensions of a tensor.</p><p>Reduces <code>input</code> along the dimensions given in <code>reduction_indices</code>. Unless
<ahref="http://docs.scipy.org/doc/numpy/user/basics.broadcasting.html">here</a></li></ul></div></div><divclass="top"><pclass="src"><aname="v:neg"class="def">neg</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: (<ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<ahref="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a><ahref="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<ahref="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a><ahref="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <ahref="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <ahref="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <ahref="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *)))))))) t)</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><tdclass="doc"><p><strong>x</strong></p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><tdclass="doc"><p><strong>y</strong></p></td></tr></table></div><divclass="doc"><p>Computes numerical negative value element-wise.</p><p>I.e., \(y = -x\).</p></div></div><divclass="top"><pclass="src"><aname="v:pack"class="def">pack</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=> [<ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t]</td><tdclass="doc"><p><strong>values</strong>: Must be of same shape and type.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><tdclass="doc"><p><strong>output</strong>: The packed tensor.</p></td></tr></table></div><divclass="doc"><p>Packs a list of <code>N</code> rank-<code>R</code> tensors into one rank-`(R+1)` tensor.</p><p>Packs the <code>N</code> tensors in <code>values</code> into a tensor with rank one higher than each
tensor in <code>values</code>, by packing them along the <code>axis</code> dimension.
Given a list of tensors of shape `(A, B, C)`;</p><p>if `axis == 0` then the <code>output</code> tensor will have the shape `(N, A, B, C)`.
if `axis == 1` then the <code>output</code> tensor will have the shape `(A, N, B, C)`.
Etc.</p><p>For example:</p><p>```prettyprint
# <code>x</code> is [1, 4]
# <code>y</code> is [2, 5]
# <code>z</code> is [3, 6]
pack([x, y, z]) => [[1, 4], [2, 5], [3, 6]] # Pack along first dim.
```</p><p>This is the opposite of <code><ahref="../tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:unpack">unpack</a></code>.</p></div></div><divclass="top"><pclass="src"><aname="v:placeholder"class="def">placeholder</a> :: <spanclass="keyword">forall</span> a. <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =><ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:Shape">Shape</a> -><ahref="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a) <ahref="src/TensorFlow-Ops.html#placeholder"class="link">Source</a></p></div><divclass="top"><pclass="src"><aname="v:range"class="def">range</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: (<ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tidx, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <ahref="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <ahref="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *))))) tidx)</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tidx</td><tdclass="doc"><p><strong>start</strong>: 0-D (scalar). First entry in the sequence.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tidx</td><tdclass="doc"><p><strong>limit</strong>: 0-D (scalar). Upper limit of sequence, exclusive.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 tidx</td><tdclass="doc"><p><strong>delta</strong>: 0-D (scalar). Optional. Default is 1. Number that increments <code>start</code>.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> tidx</td><tdclass="doc"><p><strong>output</strong>: 1-D.</p></td></tr></table></div><divclass="doc"><p>Creates a sequence of numbers.</p><p>This operation creates a sequence of numbers that begins at <code>start</code> and
the outputs of that operation (both work equivalently).</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><tdclass="doc"><p><strong>backprops</strong>: `gradients * (features > 0)`.</p></td></tr></table></div><divclass="doc"><p>Computes rectified linear gradients for a Relu operation.</p></div></div><divclass="top"><pclass="src"><aname="v:reshape"class="def">reshape</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: (<ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tshape, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) tshape)</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><tdclass="doc"><p><strong>tensor</strong></p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tshape</td><tdclass="doc"><p><strong>shape</strong>: Defines the shape of the output tensor.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><tdclass="doc"><p><strong>output</strong></p></td></tr></table></div><divclass="doc"><p>Reshapes a tensor.</p><p>Given <code>tensor</code>, this operation returns a tensor that has the same values
as <code>tensor</code> with shape <code><ahref="../tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shape">shape</a></code>.</p><p>If one component of <code><ahref="../tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shape">shape</a></code> is the special value -1, the size of that dimension
is computed so that the total size remains constant. In particular, a <code><ahref="../tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shape">shape</a></code>
of `[-1]` flattens into 1-D. At most one component of <code><ahref="../tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shape">shape</a></code> can be -1.</p><p>If <code><ahref="../tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shape">shape</a></code> is 1-D or higher, then the operation returns a tensor with shape
<code><ahref="../tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shape">shape</a></code> filled with the values of <code>tensor</code>. In this case, the number of elements
implied by <code><ahref="../tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:shape">shape</a></code> must be the same as the number of elements in <code>tensor</code>.</p><p>For example:</p><p>```prettyprint
[7, 8, 9]]</p><p># tensor <code>t</code> is [[[1, 1], [2, 2]],
# [[3, 3], [4, 4]]]
# tensor <code>t</code> has shape [2, 2, 2]
reshape(t, [2, 4]) ==> [[1, 1, 2, 2],
[3, 3, 4, 4]]</p><p># tensor <code>t</code> is [[[1, 1, 1],
# [2, 2, 2]],
# [[3, 3, 3],
# [4, 4, 4]],
# [[5, 5, 5],
# [6, 6, 6]]]
# tensor <code>t</code> has shape [3, 2, 3]
# pass '[-1]' to flatten <code>t</code>
reshape(t, [-1]) ==> [1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5, 6, 6, 6]</p><p># -1 can also be used to infer the shape</p><p># -1 is inferred to be 9:
```</p></div></div><divclass="top"><pclass="src"><aname="v:restore"class="def">restore</a><ahref="src/TensorFlow-Ops.html#restore"class="link">Source</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><tdclass="doc"><p>File path.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a</td><tdclass="doc"><p>Tensor to restore.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><tdclass="doc empty"> </td></tr></table></div><divclass="doc"><p>Restore a tensor's value from a checkpoint file.</p></div></div><divclass="top"><pclass="src"><aname="v:restoreFromName"class="def">restoreFromName</a><ahref="src/TensorFlow-Ops.html#restoreFromName"class="link">Source</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><tdclass="doc"><p>File path.</p></td></tr><tr><tdclass="src">-><ahref="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><tdclass="doc"><p>Tensor name override.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> a</td><tdclass="doc"><p>Tensor to restore.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><tdclass="doc empty"> </td></tr></table></div><divclass="doc"><p>Restore a tensor's value from a checkpoint file.</p><p>This version allows restoring from a checkpoint file that uses a different
tensor name than the variable.</p></div></div><divclass="top"><pclass="src"><aname="v:save"class="def">save</a><ahref="src/TensorFlow-Ops.html#save"class="link">Source</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../bytestring-0.10.6.0/Data-ByteString.html#t:ByteString">ByteString</a></td><tdclass="doc"><p>File path.</p></td></tr><tr><tdclass="src">-> [<ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a]</td><tdclass="doc"><p>Tensors to save.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Output.html#t:ControlNode">ControlNode</a></td><tdclass="doc empty"> </td></tr></table></div></div><divclass="top"><pclass="src"><aname="v:scalar"class="def">scalar</a> :: <spanclass="keyword">forall</span> a. <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a => a -><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a <ahref="src/TensorFlow-Ops.html#scalar"class="link">Source</a></p><divclass="doc"><p>Create a constant scalar.</p></div></div><divclass="top"><pclass="src"><aname="v:shape"class="def">shape</a> :: <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t =><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t -><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a><ahref="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a><ahref="src/TensorFlow-Ops.html#shape"class="link">Source</a></p></div><divclass="top"><pclass="src"><aname="v:sign"class="def">sign</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: (<ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * (<ahref="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a><ahref="../base-4.8.2.0/Prelude.html#t:Double">Double</a>) ((:) * (<ahref="../base-4.8.2.0/Data-Complex.html#t:Complex">Complex</a><ahref="../base-4.8.2.0/Prelude.html#t:Float">Float</a>) ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ((:) * <ahref="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <ahref="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <ahref="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *)))))))) t)</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><tdclass="doc"><p><strong>x</strong></p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><tdclass="doc"><p><strong>y</strong></p></td></tr></table></div><divclass="doc"><p>Returns an element-wise indication of the sign of a number.</p><p>`y = sign(x) = -1` if `x <ahref="0`;">0 if `x == 0`; 1 if `x</a> 0`.</p><p>For complex numbers, `y = sign(x) = x / |x|` if `x != 0`, otherwise `y = 0`.</p></div></div><divclass="top"><pclass="src"><aname="v:size"class="def">size</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: (<ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> out_type, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> (
```</p></div></div><divclass="top"><pclass="src"><aname="v:softmax"class="def">softmax</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: (<ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <ahref="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <ahref="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <ahref="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *)))) t)</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><tdclass="doc"><p><strong>logits</strong>: 2-D with shape `[batch_size, num_classes]`.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><tdclass="doc"><p><strong>softmax</strong>: Same shape as <code>logits</code>.</p></td></tr></table></div><divclass="doc"><p>Computes softmax activations.</p><p>For each batch <code>i</code> and class <code>j</code> we have</p><p>softmax[i, j] = exp(logits[i, j]) / sum_j(exp(logits[i, j]))</p></div></div><divclass="top"><pclass="src"><aname="v:softmaxCrossEntropyWithLogits"class="def">softmaxCrossEntropyWithLogits</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: (<ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <ahref="../base-4.8.2.0/Data-Word.html#t:Word16">Word16</a> ((:) * <ahref="../base-4.8.2.0/Prelude.html#t:Double">Double</a> ((:) * <ahref="../base-4.8.2.0/Prelude.html#t:Float">Float</a> ([] *)))) t)</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><tdclass="doc"><p><strong>features</strong>: batch_size x num_classes matrix</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 t</td><tdclass="doc"><p><strong>labels</strong>: batch_size x num_classes matrix
The caller must ensure that each batch of labels represents a valid
probability distribution.</p></td></tr><tr><tdclass="src">-> (<ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t, <ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t)</td><tdclass="doc"><p>(<strong>loss</strong>, <strong>backprop</strong>)</p><ul><li><strong>loss</strong>: Per example loss (batch_size vector).</li><li><strong>backprop</strong>: backpropagated gradients (batch_size x num_classes matrix).</li></ul></td></tr></table></div><divclass="doc"><p>Computes softmax cross entropy cost and gradients to backpropagate.</p><p>Inputs are the logits, not probabilities.</p></div></div><divclass="top"><pclass="src"><aname="v:sparseToDense"class="def">sparseToDense</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: (<ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tindices, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) tindices)</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 tindices</td><tdclass="doc"><p><strong>sparse_indices</strong>: 0-D, 1-D, or 2-D. `sparse_indices[i]` contains the complete
index where `sparse_values[i]` will be placed.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tindices</td><tdclass="doc"><p><strong>output_shape</strong>: 1-D. Shape of the dense output tensor.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v3 t</td><tdclass="doc"><p><strong>sparse_values</strong>: 1-D. Values corresponding to each row of <code>sparse_indices</code>,
or a scalar value to be used for all sparse indices.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v4 t</td><tdclass="doc"><p><strong>default_value</strong>: Scalar value to set for indices not specified in
<code>sparse_indices</code>.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><tdclass="doc"><p><strong>dense</strong>: Dense output tensor of shape <code>output_shape</code>.</p></td></tr></table></div><divclass="doc"><p>Converts a sparse representation into a dense tensor.</p><p>Builds an array <code>dense</code> with shape <code>output_shape</code> such that</p><p>```prettyprint
# If sparse_indices is scalar
dense[i] = (i == sparse_indices ? sparse_values : default_value)</p><p># If sparse_indices is a vector, then for each i
dense[sparse_indices[i]] = sparse_values[i]</p><p># If sparse_indices is an n by d matrix, then for each i in [0, n)
retained with length 1.</p></div></div><divclass="top"><pclass="src"><aname="v:transpose"class="def">transpose</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: (<ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> tperm, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:OneOf">OneOf</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int32">Int32</a> ((:) * <ahref="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a> ([] *))) tperm)</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><tdclass="doc"><p><strong>x</strong></p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v2 tperm</td><tdclass="doc"><p><strong>perm</strong></p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><tdclass="doc"><p><strong>y</strong></p></td></tr></table></div><divclass="doc"><p>Shuffle dimensions of x according to a permutation.</p><p>The output <code>y</code> has the same rank as <code>x</code>. The shapes of <code>x</code> and <code>y</code> satisfy:
`y.shape[i] == x.shape[perm[i]] for i in [0, 1, ..., rank(x) - 1]`</p></div></div><divclass="top"><pclass="src"><aname="v:truncatedNormal"class="def">truncatedNormal</a><ahref="src/TensorFlow-Ops.html#truncatedNormal"class="link">Source</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v <ahref="../base-4.8.2.0/Data-Int.html#t:Int64">Int64</a></td><tdclass="doc"><p>Shape.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a)</td><tdclass="doc empty"> </td></tr></table></div></div><divclass="top"><pclass="src"><aname="v:variable"class="def">variable</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> dtype</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:Shape">Shape</a></td><tdclass="doc"><p><strong>shape</strong>: The shape of the variable tensor.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Build.html#t:Build">Build</a> (<ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Ref">Ref</a> dtype)</td><tdclass="doc"><p><strong>ref</strong>: A reference to the variable tensor.</p></td></tr></table></div><divclass="doc"><p>Holds state in the form of a tensor that persists across steps.</p><p>Outputs a ref to the tensor state so it may be read or modified.
TODO(zhifengc/mrry): Adds a pointer to a more detail document
about sharing states in tensorflow.</p></div></div><divclass="top"><pclass="src"><aname="v:vector"class="def">vector</a> :: <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a => [a] -><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a <ahref="src/TensorFlow-Ops.html#vector"class="link">Source</a></p><divclass="doc"><p>Create a constant vector.</p></div></div><divclass="top"><pclass="src"><aname="v:zeros"class="def">zeros</a> :: <spanclass="keyword">forall</span> a. (<ahref="../base-4.8.2.0/Prelude.html#t:Num">Num</a> a, <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a) =><ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:Shape">Shape</a> -><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a <ahref="src/TensorFlow-Ops.html#zeros"class="link">Source</a></p></div><divclass="top"><pclass="src"><aname="v:zerosLike"class="def">zerosLike</a></p><divclass="subs arguments"><pclass="caption">Arguments</p><table><tr><tdclass="src">:: <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> t</td><tdclass="doc empty"> </td></tr><tr><tdclass="src">=><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v1 t</td><tdclass="doc"><p><strong>x</strong>: a tensor of type T.</p></td></tr><tr><tdclass="src">-><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> t</td><tdclass="doc"><p><strong>y</strong>: a tensor of the same shape and type as x but filled with zeros.</p></td></tr></table></div><divclass="doc"><p>Returns a tensor of zeros with the same shape and type as x.</p></div></div><divclass="top"><pclass="src"><aname="v:scalarize"class="def">scalarize</a> :: <ahref="../tensorflow-0.1.0.0/TensorFlow-Types.html#t:TensorType">TensorType</a> a =><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a> v a -><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Tensor">Tensor</a><ahref="../tensorflow-0.1.0.0/TensorFlow-Tensor.html#t:Value">Value</a> a <ahref="src/TensorFlow-Ops.html#scalarize"class="link">Source</a></p><divclass="doc"><p>Reshape a N-D tensor down to a scalar.</p><p>See <code><ahref="../tensorflow-core-ops-0.1.0.0/TensorFlow-GenOps-Core.html#v:reshape">reshape</a></code>.</p></div></div></div></div><divid="footer"><p>Produced by <ahref="http://www.haskell.org/haddock/">Haddock</a> version 2.16.1</p></div></body></html>
