1
0
Fork 0
mirror of https://github.com/tensorflow/haskell.git synced 2024-11-26 21:09:44 +01:00
tensorflow-haskell/tensorflow-ops/tests/EmbeddingOpsTest.hs
Judah Jacobson d62c614695 Distinguish between "rendered" and "unrendered" Tensors. (#88)
Distinguish between "rendered" and "unrendered" Tensors.

There are now three types of `Tensor`:

- `Tensor Value a`: rendered value
- `Tensor Ref a`: rendered reference
- `Tensor Build a` : unrendered value

The extra bookkeeping makes it easier to track (and enforce) which tensors are
rendered or not.  For examples where this has been confusing in the past, see

With this change, pure ops look similar to before, returning `Tensor Build`
instead of `Tensor Value`.  "Stateful" (monadic) ops are unchanged.  For
example:

    add :: OneOf [..] t => Tensor v'1 t -> Tensor v'2 t -> Tensor Build t
    assign :: (MonadBuild m, TensorType t)
           => Tensor Ref t -> Tensor v'2 t -> m (Tensor Ref t)

The `gradients` function now requires that the variables over which it's
differentiating are pre-rendered:

    gradients :: (..., Rendered v2) => Tensor v1 a -> [Tensor v2 a]
              -> m [Tensor Value a]

(`Rendered v2` means that `v2` is either a `Ref` or a `Value`.)

Additionally, the implementation of `gradients` now takes care to render every
intermediate value when performing the reverse accumulation.  I suspect this
fixes an exponential blowup for complicated expressions.
2017-04-06 15:10:33 -07:00

178 lines
7.2 KiB
Haskell

-- Copyright 2016 TensorFlow authors.
--
-- Licensed under the Apache License, Version 2.0 (the "License");
-- you may not use this file except in compliance with the License.
-- You may obtain a copy of the License at
--
-- http://www.apache.org/licenses/LICENSE-2.0
--
-- Unless required by applicable law or agreed to in writing, software
-- distributed under the License is distributed on an "AS IS" BASIS,
-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-- See the License for the specific language governing permissions and
-- limitations under the License.
{-# LANGUAGE FlexibleContexts #-}
{-# LANGUAGE RankNTypes #-}
{-# LANGUAGE ScopedTypeVariables #-}
-- | Tests for EmbeddingOps.
module Main where
import Control.Monad.IO.Class (liftIO)
import Data.Int (Int32, Int64)
import Data.List (genericLength)
import Google.Test (googleTest)
import TensorFlow.EmbeddingOps (embeddingLookup)
import Test.Framework (Test)
import Test.Framework.Providers.QuickCheck2 (testProperty)
import Test.HUnit ((@=?))
import Test.Framework.Providers.HUnit (testCase)
import Test.QuickCheck (Arbitrary(..), Property, choose, vectorOf)
import Test.QuickCheck.Monadic (monadicIO, run)
import TensorFlow.Test (assertAllClose)
import qualified Data.Vector as V
import qualified TensorFlow.GenOps.Core as CoreOps
import qualified TensorFlow.Ops as TF
import qualified TensorFlow.Session as TF
import qualified TensorFlow.Tensor as TF
import qualified TensorFlow.Types as TF
import qualified TensorFlow.Gradient as TF
import qualified TensorFlow.Build as TF
-- | Tries to perform a simple embedding lookup, with two partitions.
testEmbeddingLookupHasRightShapeWithPartition :: Test
testEmbeddingLookupHasRightShapeWithPartition =
testCase "testEmbeddingLookupHasRightShapeWithPartition" $ do
let embShape = TF.Shape [1, 3] -- Consider a 3-dim embedding of two items.
let embedding1 = [1, 1, 1 :: Int32]
let embedding2 = [0, 0, 0 :: Int32]
let idValues = [0, 1 :: Int32]
(values, shape) <- TF.runSession $ do
embedding <- mapM TF.render [ TF.constant embShape embedding1
, TF.constant embShape embedding2
]
let ids = TF.constant (TF.Shape [1, 2]) idValues
vs <- embeddingLookup embedding ids
TF.run (vs, TF.shape vs)
-- This is the shape that is returned in the equiv. Python.
shape @=? V.fromList [1, 2, 3]
-- "[0, 1]" should pull out the resulting vector.
values @=? V.fromList [1, 1, 1, 0, 0, 0]
-- | Tries to perform a simple embedding lookup, with only a single partition.
testEmbeddingLookupHasRightShape :: Test
testEmbeddingLookupHasRightShape =
testCase "testEmbeddingLookupHasRightShape" $ do
-- Consider a 3-dim embedding of two items
let embShape = TF.Shape [2, 3]
let embeddingInit = [ 1, 1, 1
, 0, 0, 0 :: Int32
]
let idValues = [0, 1 :: Int32]
(values, shape) <- TF.runSession $ do
embedding <- TF.render $ TF.constant embShape embeddingInit
let ids = TF.constant (TF.Shape [1, 2]) idValues
vs <- embeddingLookup [embedding] ids
TF.run (vs, TF.shape vs)
-- This is the shape that is returned in the equiv. Python.
shape @=? V.fromList [1, 2, 3]
-- "[0, 1]" should pull out the resulting vector.
values @=? V.fromList [1, 1, 1, 0, 0, 0]
-- | Check that we can calculate gradients w.r.t embeddings.
testEmbeddingLookupGradients :: Test
testEmbeddingLookupGradients = testCase "testEmbeddingLookupGradients" $ do
-- Agrees with "embedding", so gradient should be zero.
let xVals = V.fromList ([20, 20 :: Float])
let shape = TF.Shape [2]
gs <- TF.runSession $ do
let embShape = TF.Shape [2, 1]
let embeddingInit = [1, 20 ::Float]
let idValues = [1, 1 :: Int32]
let ids = TF.constant (TF.Shape [1, 2]) idValues
x <- TF.placeholder (TF.Shape [2])
embedding <- TF.initializedVariable
(TF.constant embShape embeddingInit)
op <- embeddingLookup [embedding] ids
let twoNorm = CoreOps.square $ TF.abs (op `TF.sub` x)
loss = TF.mean twoNorm (TF.scalar (0 :: Int32))
grad <- fmap head (TF.gradients loss [embedding])
TF.runWithFeeds
[TF.feed x $ TF.encodeTensorData shape xVals]
grad
-- Gradients should be zero (or close)
assertAllClose gs (V.fromList ([0, 0 :: Float]))
-- Verifies that direct gather is the same as dynamic split into
-- partitions, followed by embedding lookup.
testEmbeddingLookupUndoesSplit ::
forall a. (TF.TensorDataType V.Vector a, Show a, Eq a)
=> LookupExample a -> Property
testEmbeddingLookupUndoesSplit
(LookupExample numParts
shape@(TF.Shape (firstDim : restDims))
values
indices) = monadicIO $ run $ TF.runSession $ do
let shapedValues = TF.constant shape values
indicesVector <- TF.render $ TF.vector indices
let directs = CoreOps.gather shapedValues indicesVector
let cyclicCounter :: TF.Tensor TF.Build Int32 =
TF.vector [0..fromIntegral firstDim-1]
`CoreOps.mod` fromIntegral numParts
modShardedValues :: [TF.Tensor TF.Value a] <-
mapM TF.render $ CoreOps.dynamicPartition numParts shapedValues cyclicCounter
embeddings <- embeddingLookup modShardedValues indicesVector
(shapeOut, got, want :: V.Vector a) <-
TF.run (TF.cast (TF.shape embeddings), embeddings, directs)
-- Checks the explicitly documented invariant of embeddingLookup.
liftIO $ shapeOut @=? V.fromList (genericLength indices : restDims)
liftIO $ got @=? want
testEmbeddingLookupUndoesSplit _ = error "Bug in Arbitrary (LookupExample)"
-- | Consistent set of parameters for EmbeddingLookupUndoesSplit.
data LookupExample a = LookupExample
Int64 -- ^ number of ways to split.
TF.Shape -- ^ shape of the generated tensor
[a] -- ^ data for the tensor
[Int32] -- ^ indices to split the tensor by
deriving Show
instance Arbitrary a => Arbitrary (LookupExample a) where
arbitrary = do
rank <- choose (1, 4)
-- Takes rank-th root of 100 to cap the tensor size.
let maxDim = fromIntegral (ceiling doubleMaxDim :: Int64)
doubleMaxDim :: Double
doubleMaxDim = 100 ** (1 / fromIntegral rank)
shape@(firstDim : _) <- vectorOf rank (choose (1, maxDim))
values <- vectorOf (fromIntegral $ product shape) arbitrary
numParts <- choose (2, 15)
indSize <- choose (0, fromIntegral $ firstDim - 1)
indices <- vectorOf indSize (choose (0, fromIntegral firstDim - 1))
return $ LookupExample numParts (TF.Shape shape) values indices
main :: IO ()
main = googleTest
[ testProperty "EmbeddingLookupUndoesSplit"
(testEmbeddingLookupUndoesSplit :: LookupExample Double -> Property)
, testEmbeddingLookupHasRightShape
, testEmbeddingLookupHasRightShapeWithPartition
, testEmbeddingLookupGradients
]