platform/upstream/tensorflow.git
6 years agoMerge pull request #19512 from yongtang/05232018-pylint
Michael Case [Thu, 24 May 2018 01:19:24 +0000 (18:19 -0700)]
Merge pull request #19512 from yongtang/05232018-pylint

Fix pylint issue in the master branch

6 years agoAdditional pylint fix.
Yong Tang [Wed, 23 May 2018 23:53:53 +0000 (23:53 +0000)]
Additional pylint fix.

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoFix pylint issue in the master branch
Yong Tang [Wed, 23 May 2018 23:40:59 +0000 (23:40 +0000)]
Fix pylint issue in the master branch

The latest master branch has the following pylint failure that caused
`Ubuntu Sanity — Internal CI build failed`:
```
53 FAIL: Found 2 non-whitelited pylint errors:
54 tensorflow/contrib/cmake/tools/create_def_file.py:47: [C0301(line-too-long), ] Line too long (106/80)
55
56 tensorflow/contrib/cmake/tools/create_def_file.py:61: [C0301(line-too-long), ] Line too long (90/80)
```

This PR addresses the above issues.

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoExpose stream executor namespace in cmake shared object. (#19415)
Gunhan Gulsoy [Wed, 23 May 2018 22:49:35 +0000 (15:49 -0700)]
Expose stream executor namespace in cmake shared object. (#19415)

Instead of perftools::gputools expose stream executor namespace in cmake shared object.

6 years agoMerge pull request #19502 from sdalbsoo/typo
Michael Case [Wed, 23 May 2018 16:46:14 +0000 (09:46 -0700)]
Merge pull request #19502 from sdalbsoo/typo

Fix ovic typo

6 years agotypo
Sdalbsoo [Wed, 23 May 2018 15:16:54 +0000 (00:16 +0900)]
typo

6 years agoMerge pull request #17070 from yupbank/refactor-tree-variables
Martin Wicke [Tue, 22 May 2018 23:27:20 +0000 (16:27 -0700)]
Merge pull request #17070 from yupbank/refactor-tree-variables

add model restore support for tree and forest variables

6 years agoBUG: keras.callbacks.TensorBoard raises an exception for non_trainale_weights (#19148)
Yan Facai (颜发才) [Tue, 22 May 2018 23:16:01 +0000 (07:16 +0800)]
BUG: keras.callbacks.TensorBoard raises an exception for non_trainale_weights (#19148)

* TST: write_grads for non_trainable_weights

* BUG: bypass non_trainable_weights for write_grad

* CLN: factor out write_grad loop

6 years agoSetting default openmp settings for MKL kernels (#19136)
Jayaram Bobba [Tue, 22 May 2018 19:03:24 +0000 (12:03 -0700)]
Setting default openmp settings for MKL kernels (#19136)

* Change inter op defaults when built with MKL to avoid thread oversubscription

* Bump up default mkl inter_op to be less conservative

* Added default OMP settings that are expected to give reasonable performance when using MKL kernels

* Query CPUID for determining number of hyperthreads per physical core on Intel 64 architectures

* Style fixes

* Buildifier and clang-format style fixes

* Removed use of setenv due to concerns about thread safety

6 years agoMerge pull request #19386 from smistad/fix-cmake
Michael Case [Tue, 22 May 2018 18:06:19 +0000 (11:06 -0700)]
Merge pull request #19386 from smistad/fix-cmake

CMake: fix include error when building with PYTHON_BINDINGS off

6 years agoMerge pull request #19446 from sarahedkins/fix/python-feeding-function-error
Michael Case [Tue, 22 May 2018 18:02:22 +0000 (11:02 -0700)]
Merge pull request #19446 from sarahedkins/fix/python-feeding-function-error

fix(python): ValueError len(dataframe.columns) + 1

6 years agoMerge pull request #17025 from kr-ish/numpy_input_fn_none_check
Michael Case [Tue, 22 May 2018 16:59:54 +0000 (09:59 -0700)]
Merge pull request #17025 from kr-ish/numpy_input_fn_none_check

Adds a check for shuffle=None in numpy_input_fn

6 years agoMerge pull request #19445 from yongtang/05122018-link
Michael Case [Tue, 22 May 2018 16:56:49 +0000 (09:56 -0700)]
Merge pull request #19445 from yongtang/05122018-link

Fix incorrect link for nvidia drivers

6 years agoMerge pull request #19462 from yongtang/05212018-docs-math_ops
Michael Case [Tue, 22 May 2018 16:51:02 +0000 (09:51 -0700)]
Merge pull request #19462 from yongtang/05212018-docs-math_ops

Fix discrepancies between doc and implementation for math_ops

6 years agoFix typo (#19450)
ManHyuk [Tue, 22 May 2018 13:43:40 +0000 (22:43 +0900)]
Fix typo (#19450)

6 years agofix(python): ValueError len(dataframe.columns) + 1
Sarah Edkins [Tue, 22 May 2018 00:24:40 +0000 (20:24 -0400)]
fix(python): ValueError len(dataframe.columns) + 1

6 years agoUpdate docstring for tf.round as it support all real types.
Yong Tang [Mon, 21 May 2018 23:57:17 +0000 (23:57 +0000)]
Update docstring for tf.round as it support all real types.

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoUpdate docstring as pow supports `float16` now
Yong Tang [Mon, 21 May 2018 23:36:55 +0000 (23:36 +0000)]
Update docstring as pow supports `float16` now

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoFix discrepancies between doc and implementation for math_ops
Yong Tang [Mon, 21 May 2018 23:29:32 +0000 (23:29 +0000)]
Fix discrepancies between doc and implementation for math_ops

In the docstring of math_ops, there are some discrepancies
between the supported data types and the actual implementation:
`abs` supports `float16`.

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoMerge pull request #19254 from yongtang/05122018-compute_accidental_hits
Michael Case [Mon, 21 May 2018 22:48:11 +0000 (15:48 -0700)]
Merge pull request #19254 from yongtang/05122018-compute_accidental_hits

Add additional shape validation to `compute_accidental_hits`

6 years agoMerge pull request #19257 from yongtang/05122018-QuantizedReluX
Michael Case [Mon, 21 May 2018 21:38:18 +0000 (14:38 -0700)]
Merge pull request #19257 from yongtang/05122018-QuantizedReluX

Shape validation of `max_features` in `QuantizedReluX`

6 years agoMerge pull request #19326 from yongtang/05162018-mpi_utils.h
Xiaoqiang Zheng [Mon, 21 May 2018 19:40:59 +0000 (12:40 -0700)]
Merge pull request #19326 from yongtang/05162018-mpi_utils.h

Remove redundant header includes in mpi_utils.h

6 years agoaddress lint again
Peng Yu [Mon, 21 May 2018 01:20:46 +0000 (21:20 -0400)]
address lint again

6 years agoaddress lint
Peng Yu [Fri, 18 May 2018 21:05:45 +0000 (17:05 -0400)]
address lint

6 years agoaddress comments
Peng Yu [Thu, 12 Apr 2018 19:38:48 +0000 (15:38 -0400)]
address comments

6 years agoadd test case always predict [0, 1]
Peng Yu [Thu, 12 Apr 2018 02:04:58 +0000 (22:04 -0400)]
add test case always predict [0, 1]

6 years agoadd inference support for tree and forest variables
Peng Yu [Fri, 16 Feb 2018 15:59:14 +0000 (10:59 -0500)]
add inference support for tree and forest variables

6 years agoMerge pull request #19258 from Androbin/patch-2
Michael Case [Mon, 21 May 2018 17:30:43 +0000 (10:30 -0700)]
Merge pull request #19258 from Androbin/patch-2

Update TFLite Docs on tf.gather

6 years agoMerge pull request #19394 from miaout17/fix-demop
Michael Case [Mon, 21 May 2018 17:27:14 +0000 (10:27 -0700)]
Merge pull request #19394 from miaout17/fix-demop

Fix the TFLite iOS demo apps.

6 years agoMerge pull request #19396 from yongtang/19338-clip
Michael Case [Mon, 21 May 2018 17:26:42 +0000 (10:26 -0700)]
Merge pull request #19396 from yongtang/19338-clip

Add test case for empty tensor with clip ops

6 years agoMerge pull request #19432 from soonson/patch-1
Michael Case [Mon, 21 May 2018 17:10:03 +0000 (10:10 -0700)]
Merge pull request #19432 from soonson/patch-1

Update groups.md

6 years agoDOC: Fix python code in for invalid code (#19218)
Abdullah Alrasheed [Mon, 21 May 2018 16:08:27 +0000 (00:08 +0800)]
DOC: Fix python code in for invalid code (#19218)

There is an error in python code in the documentation. There is no class called `tf.estimator.Estimator.LinearClassifier` I think it was a typo and the author meant `tf.estimator.LinearClassifier`.

6 years agoUpdate groups.md
soonson [Mon, 21 May 2018 11:14:54 +0000 (20:14 +0900)]
Update groups.md

Add more user groups collected

6 years agoMerge pull request #19357 from yongtang/03122018-_any_variable_initialized
Xiaoqiang Zheng [Mon, 21 May 2018 07:24:14 +0000 (00:24 -0700)]
Merge pull request #19357 from yongtang/03122018-_any_variable_initialized

Fix the naming of _any_variable_initialized

6 years agoMerge pull request #19273 from ageron/fix_estimator_doc
Xiaoqiang Zheng [Mon, 21 May 2018 06:45:37 +0000 (23:45 -0700)]
Merge pull request #19273 from ageron/fix_estimator_doc

Fix errors and typos in the Estimators programmer's guide

6 years agoStreamingFilesDataset fixes (#19413)
Nand Dalal [Mon, 21 May 2018 03:15:21 +0000 (22:15 -0500)]
StreamingFilesDataset fixes (#19413)

* use source_dataset.output_dtypes to yield correctly typed output dataset

* add test and fix issue introduced by 2a6c5998a239f41926ca295ac20bb595862fd5ff

6 years agoFix typo (#19411)
ManHyuk [Sun, 20 May 2018 14:48:05 +0000 (23:48 +0900)]
Fix typo (#19411)

6 years agoMerge pull request #18685 from nuchi/fft_no_core_framework_and_single_thread
Martin Wicke [Sun, 20 May 2018 00:09:51 +0000 (17:09 -0700)]
Merge pull request #18685 from nuchi/fft_no_core_framework_and_single_thread

XLA CPU Fft no core framework and single thread

6 years agocomment typo (#19381)
Qing ZHao [Sat, 19 May 2018 14:45:29 +0000 (22:45 +0800)]
comment typo (#19381)

6 years agoMerge pull request #19409 from zheng-xq/branch_197218170
Gunhan Gulsoy [Sat, 19 May 2018 03:30:46 +0000 (20:30 -0700)]
Merge pull request #19409 from zheng-xq/branch_197218170

Branch 197218170

6 years agoMerge pull request #19408 from annarev/add_more_missing_deps
annarev [Sat, 19 May 2018 02:10:04 +0000 (19:10 -0700)]
Merge pull request #19408 from annarev/add_more_missing_deps

Add missing dependencies to test_lite_main

6 years agoMerge commit for internal changes
zhengxq [Sat, 19 May 2018 00:05:40 +0000 (17:05 -0700)]
Merge commit for internal changes

6 years agoMerge pull request #19338 from girving/clip
Xiaoqiang Zheng [Fri, 18 May 2018 23:37:27 +0000 (16:37 -0700)]
Merge pull request #19338 from girving/clip

Make tf.clip_by_value not crash on empty tensors

6 years agoAutomated g4 rollback of changelist 197070234
Igor Saprykin [Fri, 18 May 2018 23:33:19 +0000 (16:33 -0700)]
Automated g4 rollback of changelist 197070234

PiperOrigin-RevId: 197218170

6 years agoImprove import error messages.
A. Unique TensorFlower [Fri, 18 May 2018 23:28:59 +0000 (16:28 -0700)]
Improve import error messages.

PiperOrigin-RevId: 197217638

6 years agoAdd missing dependencies to test_lite_main
Anna R [Fri, 18 May 2018 23:11:30 +0000 (16:11 -0700)]
Add missing dependencies to test_lite_main

6 years agoSkip tests that require unavailable hardware when not using DistributionStrategy
Igor Saprykin [Fri, 18 May 2018 22:49:01 +0000 (15:49 -0700)]
Skip tests that require unavailable hardware when not using DistributionStrategy

Right now combinations.py skips tests that do not have the hardware that's
requried by the DistributionStrategy instance that is used in that test.  After
this change, the user can trigger such a behavior even when they are not using
DistributionStrategy.

Two new special arguments are added: "required_tpu" and "required_gpus".

PiperOrigin-RevId: 197212466

6 years agoMerge pull request #19404 from yifeif/distributed
Yifei Feng [Fri, 18 May 2018 22:43:30 +0000 (15:43 -0700)]
Merge pull request #19404 from yifeif/distributed

Make distributed_test use tf-nightly by default.

6 years agoCorrect dtype in resource_strided_slice_assign
Alexandre Passos [Fri, 18 May 2018 22:33:00 +0000 (15:33 -0700)]
Correct dtype in resource_strided_slice_assign

PiperOrigin-RevId: 197210273

6 years agoRemove unused BUILD dependencies
A. Unique TensorFlower [Fri, 18 May 2018 22:15:06 +0000 (15:15 -0700)]
Remove unused BUILD dependencies

PiperOrigin-RevId: 197207799

6 years agoFixed an issue when add context into params.
Jianwei Xie [Fri, 18 May 2018 21:59:01 +0000 (14:59 -0700)]
Fixed an issue when add context into params.

PiperOrigin-RevId: 197205327

6 years agoMake distributed_test use tf-nightly by default.
Yifei Feng [Fri, 18 May 2018 21:25:59 +0000 (14:25 -0700)]
Make distributed_test use tf-nightly by default.

6 years agoAdding back abi and stacktrace dependencies to stacktrace_handler (#19401)
annarev [Fri, 18 May 2018 21:12:26 +0000 (14:12 -0700)]
Adding back abi and stacktrace dependencies to stacktrace_handler (#19401)

6 years agoRevert a change to fix TFLite iOS demo app.
Yu-Cheng Ling [Fri, 18 May 2018 20:12:41 +0000 (13:12 -0700)]
Revert a change to fix TFLite iOS demo app.
It depends on released CocoaPod.

PiperOrigin-RevId: 197189734

6 years agoThe quantizer should match the patterns for partition variables.
Suharsh Sivakumar [Fri, 18 May 2018 20:07:18 +0000 (13:07 -0700)]
The quantizer should match the patterns for partition variables.

PiperOrigin-RevId: 197189118

6 years ago[dataset]: Remove extra `repeat` in the docstring for `shard`.
A. Unique TensorFlower [Fri, 18 May 2018 19:41:20 +0000 (12:41 -0700)]
[dataset]: Remove extra `repeat` in the docstring for `shard`.

PiperOrigin-RevId: 197185877

6 years ago[TF:XLA] Delete the reverseDFS scheduler. With recent improvements to the List schedu...
Dimitris Vardoulakis [Fri, 18 May 2018 19:24:20 +0000 (12:24 -0700)]
[TF:XLA] Delete the reverseDFS scheduler. With recent improvements to the List scheduler, reverseDFS is no longer needed.

PiperOrigin-RevId: 197183727

6 years agoExpose read-only versions of input tensors in tflite.
A. Unique TensorFlower [Fri, 18 May 2018 19:21:44 +0000 (12:21 -0700)]
Expose read-only versions of input tensors in tflite.

PiperOrigin-RevId: 197183345

6 years agoEnable `SeedStream` construction from other `SeedStream` instances.
Joshua V. Dillon [Fri, 18 May 2018 19:17:05 +0000 (12:17 -0700)]
Enable `SeedStream` construction from other `SeedStream` instances.

PiperOrigin-RevId: 197182686

6 years agoAdd eager tests for automatic tensor flattening
Igor Ganichev [Fri, 18 May 2018 19:14:47 +0000 (12:14 -0700)]
Add eager tests for automatic tensor flattening

PiperOrigin-RevId: 197182339

6 years agoAdd an in-memory evaluator for Estimator. It will run evaluation without a checkpoint...
Mustafa Ispir [Fri, 18 May 2018 19:09:55 +0000 (12:09 -0700)]
Add an in-memory evaluator for Estimator. It will run evaluation without a checkpoint. This will let user get evaluation metrics on more steps without saving.

  Example:

  ```python
  def train_input_fn():
    ...
    return train_dataset

  def eval_input_fn():
    ...
    return eval_dataset

  estimator = tf.estimator.DNNClassifier(...)

  evaluator = tf.contrib.estimator.InMemoryEvaluatorHook(
      estimator, eval_input_fn)
  estimator.train(train_input_fn, hooks=[evaluator])
  ```

PiperOrigin-RevId: 197181726

6 years agoWhen converting an unsupported op in TOCO, we can possibly infer the output type...
A. Unique TensorFlower [Fri, 18 May 2018 19:08:32 +0000 (12:08 -0700)]
When converting an unsupported op in TOCO, we can possibly infer the output type from an attribute named Tout (this works, e.g., for RandomUniformInt).

PiperOrigin-RevId: 197181552

6 years ago[TF:XLA] Bump open source llvm revision to r332697
Sanjoy Das [Fri, 18 May 2018 18:56:28 +0000 (11:56 -0700)]
[TF:XLA] Bump open source llvm revision to r332697

PiperOrigin-RevId: 197179581

6 years agoboosted_trees: allow labels to have rank 1 as other estimators.
Younghee Kwon [Fri, 18 May 2018 18:46:54 +0000 (11:46 -0700)]
boosted_trees: allow labels to have rank 1 as other estimators.

PiperOrigin-RevId: 197178169

6 years agoFix the TFLite iOS demo apps.
Yu-Cheng Ling [Fri, 18 May 2018 18:46:06 +0000 (11:46 -0700)]
Fix the TFLite iOS demo apps.

This partially reverted 2bf2308872a6dfa8d6d0809acf0098f666e00fe8.
The demo apps in `lite/examples` depends on CocoaPod, not the Github
head code.

6 years ago[TF:XLA] Remove underscore prefix from XlaLaunch operator.
Peter Hawkins [Fri, 18 May 2018 18:43:25 +0000 (11:43 -0700)]
[TF:XLA] Remove underscore prefix from XlaLaunch operator.
Minor fixes to comments.

PiperOrigin-RevId: 197177582

6 years agoValidate the file signatures before issusing a content read in GCS. Remove the
Ruoxin Sang [Fri, 18 May 2018 18:28:18 +0000 (11:28 -0700)]
Validate the file signatures before issusing a content read in GCS. Remove the
checkpoint hack. Enable the stat cache by default, as the block cache has
already been enabled by default. Set the default stat cache max_age to 5s,
which is the same in the Cloud TPU server.

PiperOrigin-RevId: 197175258

6 years agoUpdate test case to cover all scenarios
Yong Tang [Fri, 18 May 2018 17:59:31 +0000 (17:59 +0000)]
Update test case to cover all scenarios

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoAdd additional test case
Yong Tang [Fri, 18 May 2018 17:58:57 +0000 (17:58 +0000)]
Add additional test case

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoAdd test case for empty tensor with clip ops.
Yong Tang [Fri, 18 May 2018 17:29:26 +0000 (17:29 +0000)]
Add test case for empty tensor with clip ops.

This fix is based on 19337 and 19338. The issue was that
previously an empty tensor for tf.clip_by_value on GPU
triggers a crash. The issue should have been fixed by 19338 and
the recent master. It makes sense to adds the test case for this issue.

This fix adds the test case.

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoAutomated g4 rollback of changelist 197118212
A. Unique TensorFlower [Fri, 18 May 2018 17:42:50 +0000 (10:42 -0700)]
Automated g4 rollback of changelist 197118212

PiperOrigin-RevId: 197167501

6 years agoOp version: Populate version in Toco TFLite exporter
Yu-Cheng Ling [Fri, 18 May 2018 17:39:45 +0000 (10:39 -0700)]
Op version: Populate version in Toco TFLite exporter

PiperOrigin-RevId: 197166962

6 years ago[TF:XLA] Allow the HloEvaluator to be changed for the HLO interpreter.
Nick Desaulniers [Fri, 18 May 2018 17:23:34 +0000 (10:23 -0700)]
[TF:XLA] Allow the HloEvaluator to be changed for the HLO interpreter.

PiperOrigin-RevId: 197164443

6 years ago[tf.data] Changed internal implementation of `make_csv_dataset`, and removed argument...
Rachel Lim [Fri, 18 May 2018 17:21:59 +0000 (10:21 -0700)]
[tf.data] Changed internal implementation of `make_csv_dataset`, and removed arguments `default_float_type` and `comment` from `make_csv_dataset`

PiperOrigin-RevId: 197164167

6 years agoImprove TFLite generated example test speed.
Yu-Cheng Ling [Fri, 18 May 2018 17:12:37 +0000 (10:12 -0700)]
Improve TFLite generated example test speed.

PiperOrigin-RevId: 197162686

6 years agoUpdate the docstring on function.Defun to reflect current graph/session freezing...
Brian Patton [Fri, 18 May 2018 15:38:40 +0000 (08:38 -0700)]
Update the docstring on function.Defun to reflect current graph/session freezing behavior.

PiperOrigin-RevId: 197150790

6 years agoRestructure BUILD rules to be modularly correct
A. Unique TensorFlower [Fri, 18 May 2018 14:52:32 +0000 (07:52 -0700)]
Restructure BUILD rules to be modularly correct

PiperOrigin-RevId: 197145545

6 years agoMark link-time dependency library as alwayslink=1.
A. Unique TensorFlower [Fri, 18 May 2018 14:49:06 +0000 (07:49 -0700)]
Mark link-time dependency library as alwayslink=1.

PiperOrigin-RevId: 197145205

6 years agoMerge pull request #18240 from Intel-tensorflow/ashraf/direct_session_bugfix
Tatiana Shpeisman [Fri, 18 May 2018 14:50:54 +0000 (07:50 -0700)]
Merge pull request #18240 from Intel-tensorflow/ashraf/direct_session_bugfix

INTEL MKL:  Fixing a unit test failure, for direct session memory allocation

6 years agoModify PadInsertion pass so that it matches other passes.
A. Unique TensorFlower [Fri, 18 May 2018 14:47:41 +0000 (07:47 -0700)]
Modify PadInsertion pass so that it matches other passes.

Currently, PadInsertion only iterates over the instructions in the
entry_computation. Other passes iterate over MakeNonfusionComputations.
When we run on HloSnapshots derived from TPU benchmarks, this makes a
difference, because it seems none of the convolutions are inside the entry
computation.

PiperOrigin-RevId: 197145067

6 years agofix include error when building with PYTHON_BINDINGS off
Erik Smistad [Fri, 18 May 2018 14:08:39 +0000 (16:08 +0200)]
fix include error when building with PYTHON_BINDINGS off

6 years agoDropping support for CUDA < 8.
A. Unique TensorFlower [Fri, 18 May 2018 13:31:20 +0000 (06:31 -0700)]
Dropping support for CUDA < 8.

PiperOrigin-RevId: 197137612

6 years agoExtracts the following optimizations into methods:
A. Unique TensorFlower [Fri, 18 May 2018 13:27:13 +0000 (06:27 -0700)]
Extracts the following optimizations into methods:

SimplifyArithmeticOperations
ReduceDivToReciprocalMul

PiperOrigin-RevId: 197137281

6 years agoUpdate downloadable clang to a new revision
Ilya Biryukov [Fri, 18 May 2018 10:14:27 +0000 (03:14 -0700)]
Update downloadable clang to a new revision

PiperOrigin-RevId: 197121447

6 years agoDropping support for cuDNN < 6.
A. Unique TensorFlower [Fri, 18 May 2018 09:37:52 +0000 (02:37 -0700)]
Dropping support for cuDNN < 6.
Enable CUDNN_FFT_TILING_FORWARD for cuDNN >= 7.

PiperOrigin-RevId: 197118212

6 years agoMove Keras code out of _impl folder and remove API files.
Pavithra Vijay [Fri, 18 May 2018 04:36:39 +0000 (21:36 -0700)]
Move Keras code out of _impl folder and remove API files.

PiperOrigin-RevId: 197097430

6 years ago[XLA] Clamp indices in DynamicSlice and DynamicUpdateSlice instead of wrapping.
Michael Kuperstein [Fri, 18 May 2018 01:09:45 +0000 (18:09 -0700)]
[XLA] Clamp indices in DynamicSlice and DynamicUpdateSlice instead of wrapping.

This implements the following index clamping in all backends (CPU, GPU, Interpreter):

for(int i = 0; i < rank; ++i)
  start_index[i] = clamp(start_index[i], 0, output_dim_size[i] - update_dim_size[i])

Which ensures the slice (or update region) is always inbounds w.r.t the input.

PiperOrigin-RevId: 197082276

6 years agoHandle HloInstruction::ToString() when literal is missing.
A. Unique TensorFlower [Fri, 18 May 2018 00:40:47 +0000 (17:40 -0700)]
Handle HloInstruction::ToString() when literal is missing.

PiperOrigin-RevId: 197079144

6 years agomarkdown fixes
Mark Daoust [Fri, 18 May 2018 00:25:47 +0000 (17:25 -0700)]
markdown fixes

PiperOrigin-RevId: 197077588

6 years agoDo not force default layout when there is no need to.
A. Unique TensorFlower [Fri, 18 May 2018 00:11:47 +0000 (17:11 -0700)]
Do not force default layout when there is no need to.
Allow the inner computations to negotiate a root and parameter layouts different from default.
END_PUBLIC

RELNOTES: n/a

---------------------
BEGIN_PUBLIC
Automated g4 rollback of changelist 194293187

PiperOrigin-RevId: 197076025

6 years agoMerge pull request #19355 from tensorflow/mrry-patch-2
Martin Wicke [Fri, 18 May 2018 00:06:37 +0000 (17:06 -0700)]
Merge pull request #19355 from tensorflow/mrry-patch-2

Include <algorithm> in .pb_text.cc generated files

6 years ago Add NCHW_VECT_W tensor format.
A. Unique TensorFlower [Thu, 17 May 2018 23:59:50 +0000 (16:59 -0700)]
  Add NCHW_VECT_W tensor format.

PiperOrigin-RevId: 197074411

6 years agoEnable optimizations submitted during pre-NIPS freeze:
A. Unique TensorFlower [Thu, 17 May 2018 23:52:31 +0000 (16:52 -0700)]
Enable optimizations submitted during pre-NIPS freeze:
  * Snapshot to Identity promotion.
  * Removal of transposes pairs separated by chain.

PiperOrigin-RevId: 197073602

6 years agoboosted_trees: fixed the crash when eval/prediction is attempted with the initial...
Younghee Kwon [Thu, 17 May 2018 23:52:17 +0000 (16:52 -0700)]
boosted_trees: fixed the crash when eval/prediction is attempted with the initial checkpoints (at step 0) before training.

PiperOrigin-RevId: 197073582

6 years agoMisc changes found when debugging an unrelated problem.
Akshay Modi [Thu, 17 May 2018 23:52:07 +0000 (16:52 -0700)]
Misc changes found when debugging an unrelated problem.

- (pywrap_tfe.i) Improve error message for better debugging TFE_Py_Execute failures.
- (pywrap_tfe_src.cc) Accept _value of None
- (base.i) Remove unnecessary temporary

PiperOrigin-RevId: 197073571

6 years agoMake V1 metrics distributed-aware. Also fix a bug where assertAllClose
A. Unique TensorFlower [Thu, 17 May 2018 23:26:24 +0000 (16:26 -0700)]
Make V1 metrics distributed-aware. Also fix a bug where assertAllClose
was sometimes ignoring its `msg` parameter.

PiperOrigin-RevId: 197070234

6 years ago[XLA] Redesign: delete xla::Computation.
A. Unique TensorFlower [Thu, 17 May 2018 23:23:33 +0000 (16:23 -0700)]
[XLA] Redesign: delete xla::Computation.

PiperOrigin-RevId: 197069851

6 years agoDon't add branch prediction hints when compiling under nvcc.
Craig Citro [Thu, 17 May 2018 23:06:21 +0000 (16:06 -0700)]
Don't add branch prediction hints when compiling under nvcc.

As seen in #19203, the `__builtin_expect` compiler builtin isn't recognized as
a builtin in nvcc8, leading to compilation failures of the form

    ./tensorflow/core/kernels/gather_functor_gpu.cu.h(57): error: calling a __host__ function("__builtin_
expect") from a __global__ function("tensorflow::GatherOpKernel< ::Eigen::half, int, (bool)1> ") is n
ot allowed

when attempting to build TensorFlow.

This change fixes things by adding an additional check for `__NVCC__`, and
avoiding any branch prediction hints in that case.

PiperOrigin-RevId: 197067418

6 years agofixing some incorrect documentation in test cases.
Olivia Nordquist [Thu, 17 May 2018 22:49:12 +0000 (15:49 -0700)]
fixing some incorrect documentation in test cases.

PiperOrigin-RevId: 197064874

6 years agoIntroduce an "indexed array" analysis
Sanjoy Das [Thu, 17 May 2018 22:47:30 +0000 (15:47 -0700)]
Introduce an "indexed array" analysis

Context: we want to optimize computations hanging off of a embedding lookup from
a constant array.  For instance, consider:

  embedding = gather from a constant array using non-constant indices
  embedding_reshaped = reshape embedding
  embedding_reshaped_transposed = transpose embedding_reshaped
  result = dot(embedding_reshaped_transposed, constant)

In the graph above, depending on how the details work out, we may be able to
fold `result` into a gather from a precomputed constant array.  However, it is
inconvenient to get there by incremental rewrites -- it is probably not
profitable to rewrite embedding_reshaped or embedding_reshaped_transposed [0] as
embedding lookups but we get to "see" that the dot can be rewritten only after
rewriting the reshape and the transpose.

This analysis aims to make the optimization above more straightforward by
allowing a transformation pass (that uses this analysis) to query the analysis
to see if if `result` _can_ be represented as an embedding lookup.  If yes it
can then apply some profitability heuristics to decide if it is worth it to
rewrite it as one.  This suggested workflow gives us separation of concerns (the
legality of the rewrite is computed separately from its profitability) and, more
importantly, lets us "look ahead" and analyze the dot without rewriting its
operands.

The implementation is far from complete (most of the interesting bits are TODO)
but I wanted to get an early design review before I spent too much time on this.

[0] Under the assumption that transposing or reshaping are not expensive enough
to pay the price of keeping around a new potentially large constant (in
particular, some of these may have been equivalent to free bitcasts).

PiperOrigin-RevId: 197064648