platform/upstream/tensorflow.git
6 years agoMerge pull request #18843 from av8ramit/fix_devel_docker
Amit Patankar [Wed, 25 Apr 2018 17:04:03 +0000 (10:04 -0700)]
Merge pull request #18843 from av8ramit/fix_devel_docker

Fixing the mock import error for devel docker.

6 years agoFixing the mock import error for devel docker.
Amit Patankar [Tue, 24 Apr 2018 23:34:01 +0000 (16:34 -0700)]
Fixing the mock import error for devel docker.

6 years agoMerge pull request #18817 from lamberta/fix-install-linux
Billy Lamberta [Tue, 24 Apr 2018 22:48:08 +0000 (15:48 -0700)]
Merge pull request #18817 from lamberta/fix-install-linux

Docs: Clean up install_linux page

6 years agodocs: install_linux, move GPU section below install procedures.
Billy Lamberta [Tue, 24 Apr 2018 21:52:38 +0000 (14:52 -0700)]
docs: install_linux, move GPU section below install procedures.

6 years agoMerge pull request #18836 from tensorflow/av8ramit-patch-3
Martin Wicke [Tue, 24 Apr 2018 19:16:18 +0000 (12:16 -0700)]
Merge pull request #18836 from tensorflow/av8ramit-patch-3

Update README.md

6 years agoReview fixes to install_linux
Billy Lamberta [Tue, 24 Apr 2018 18:09:09 +0000 (11:09 -0700)]
Review fixes to install_linux

6 years agoUpdate README.md
Amit Patankar [Tue, 24 Apr 2018 17:59:10 +0000 (10:59 -0700)]
Update README.md

6 years agodocs: Clean up install_linux with pip
Billy Lamberta [Tue, 24 Apr 2018 05:08:52 +0000 (22:08 -0700)]
docs: Clean up install_linux with pip

6 years agoMerge pull request #18804 from tensorflow/merge_branch
Yifei Feng [Mon, 23 Apr 2018 22:22:57 +0000 (15:22 -0700)]
Merge pull request #18804 from tensorflow/merge_branch

R1.8 merge back to master after 1.8.0-rc1

6 years agoReapply #18446.
Yifei Feng [Mon, 23 Apr 2018 21:24:11 +0000 (14:24 -0700)]
Reapply #18446.

6 years agoFix tfcompile module label. (#16582)
Andy Kernahan [Mon, 23 Apr 2018 19:32:35 +0000 (20:32 +0100)]
Fix tfcompile module label. (#16582)

6 years agoUpdate resources.h
Yifei Feng [Mon, 23 Apr 2018 18:44:22 +0000 (11:44 -0700)]
Update resources.h

6 years agoMerge remote-tracking branch 'origin/r1.8' into merge_branch
Yifei Feng [Mon, 23 Apr 2018 18:40:55 +0000 (11:40 -0700)]
Merge remote-tracking branch 'origin/r1.8' into merge_branch

6 years agoMerge branch 'master' into r1.8
Yifei Feng [Mon, 23 Apr 2018 17:53:11 +0000 (10:53 -0700)]
Merge branch 'master' into r1.8

6 years agoMerge pull request #18746 from yifeif/branch_193740595
Yifei Feng [Mon, 23 Apr 2018 17:02:49 +0000 (10:02 -0700)]
Merge pull request #18746 from yifeif/branch_193740595

Branch 193740595

6 years agoMerge pull request #18765 from Intel-tensorflow/agramesh/String_fix
Martin Wicke [Mon, 23 Apr 2018 15:41:14 +0000 (08:41 -0700)]
Merge pull request #18765 from Intel-tensorflow/agramesh/String_fix

[INTEL MKL] Fix compilation failure.

6 years agoMerge pull request #17556 from kbsriram/strided-slice-grad
Martin Wicke [Mon, 23 Apr 2018 06:53:22 +0000 (23:53 -0700)]
Merge pull request #17556 from kbsriram/strided-slice-grad

C++ gradient for StridedSlice

6 years agoMerge pull request #15354 from facaiy/ENH/scatter_nd_supports_string
Martin Wicke [Mon, 23 Apr 2018 06:52:55 +0000 (23:52 -0700)]
Merge pull request #15354 from facaiy/ENH/scatter_nd_supports_string

CPU: Support for DT_STRING type in ScatterNd

6 years agoMerge pull request #18754 from ADiegoCAlonso/fix-numerical-warning
Martin Wicke [Mon, 23 Apr 2018 06:52:38 +0000 (23:52 -0700)]
Merge pull request #18754 from ADiegoCAlonso/fix-numerical-warning

Fix numerical warning in the Iris example

6 years agoMerge pull request #18728 from jerry4897/master
Martin Wicke [Mon, 23 Apr 2018 06:52:23 +0000 (23:52 -0700)]
Merge pull request #18728 from jerry4897/master

Typo?

6 years agoMerge pull request #18698 from imsheridan/fix_api_rendering
Martin Wicke [Mon, 23 Apr 2018 06:52:08 +0000 (23:52 -0700)]
Merge pull request #18698 from imsheridan/fix_api_rendering

Fix code block rendering in several api definitions

6 years agoMerge pull request #18715 from yongtang/18712-tf.count_nonzero
Martin Wicke [Mon, 23 Apr 2018 06:51:41 +0000 (23:51 -0700)]
Merge pull request #18715 from yongtang/18712-tf.count_nonzero

Support string tensors for `tf.count_nonzero`

6 years agoMerge pull request #18753 from ADiegoCAlonso/fix-tutorials-directories-are-not-python...
Martin Wicke [Mon, 23 Apr 2018 06:48:54 +0000 (23:48 -0700)]
Merge pull request #18753 from ADiegoCAlonso/fix-tutorials-directories-are-not-python-packages

Fix tutorials directories are not Python packages

6 years agoMerge pull request #18766 from hedenface/patch-1
Martin Wicke [Mon, 23 Apr 2018 03:36:38 +0000 (20:36 -0700)]
Merge pull request #18766 from hedenface/patch-1

update $ source spacing

6 years agoMerge pull request #18778 from yongtang/04222018-squeeze_dims
Martin Wicke [Mon, 23 Apr 2018 03:33:36 +0000 (20:33 -0700)]
Merge pull request #18778 from yongtang/04222018-squeeze_dims

Change from squeeze_dims to axis when calling tf.squeeze

6 years agoMerge pull request #18760 from yongtang/04212018-gitignore
Martin Wicke [Mon, 23 Apr 2018 03:33:00 +0000 (20:33 -0700)]
Merge pull request #18760 from yongtang/04212018-gitignore

Update .gitignore for cmake generated files

6 years agoFix more for cuda version check.
Yifei Feng [Mon, 23 Apr 2018 03:00:54 +0000 (20:00 -0700)]
Fix more for cuda version check.

6 years agoUpdate tuple for cuda version with auto as it was removed in #18434.
Yifei Feng [Mon, 23 Apr 2018 02:08:21 +0000 (19:08 -0700)]
Update tuple for cuda version with auto as it was removed in #18434.

6 years agoFix warnings in tf.contrib.distributions with squeeze_dims
Yong Tang [Sun, 22 Apr 2018 17:57:31 +0000 (17:57 +0000)]
Fix warnings in tf.contrib.distributions with squeeze_dims

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoFix warnings in tf.contrib.factorization
Yong Tang [Sun, 22 Apr 2018 17:57:06 +0000 (17:57 +0000)]
Fix warnings in tf.contrib.factorization

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoFix tf.contrib.layers warnings where squeeze_dims were used with tf.squeeze
Yong Tang [Sun, 22 Apr 2018 17:56:09 +0000 (17:56 +0000)]
Fix tf.contrib.layers warnings where squeeze_dims were used with tf.squeeze

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoFix squeeze_dims warnings in tf.contrib.learn
Yong Tang [Sun, 22 Apr 2018 17:55:35 +0000 (17:55 +0000)]
Fix squeeze_dims warnings in tf.contrib.learn

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoFix warnings in tf.contrib.tensor_forest
Yong Tang [Sun, 22 Apr 2018 17:55:06 +0000 (17:55 +0000)]
Fix warnings in tf.contrib.tensor_forest

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoFix tf.contrib.timeseries warnings related to squeeze_dims
Yong Tang [Sun, 22 Apr 2018 17:54:26 +0000 (17:54 +0000)]
Fix tf.contrib.timeseries warnings related to squeeze_dims

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoFix warnings in reduce_join_op_test.py
Yong Tang [Sun, 22 Apr 2018 17:53:58 +0000 (17:53 +0000)]
Fix warnings in reduce_join_op_test.py

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoFix warning in tf.nn ops where squeeze_dims was used with tf.squeeze
Yong Tang [Sun, 22 Apr 2018 17:53:10 +0000 (17:53 +0000)]
Fix warning in tf.nn ops where squeeze_dims was used with tf.squeeze

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoFix warning in image related ops.
Yong Tang [Sun, 22 Apr 2018 17:52:31 +0000 (17:52 +0000)]
Fix warning in image related ops.

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoChange from squeeze_dims to axis when calling tf.squeeze
Yong Tang [Sun, 22 Apr 2018 17:49:13 +0000 (17:49 +0000)]
Change from squeeze_dims to axis when calling tf.squeeze

The `squeeze_dims` in `tf.squeeze` has been deprecated in favor
of `axis` while many places still use `squeeze_dims`. That
generates lots of warnings.

This fix switches from `squeeze_dims` to `axis` to remove those warnings.

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoUpdate docs to add note and examples for tf.count_nonzero with string
Yong Tang [Sun, 22 Apr 2018 14:43:21 +0000 (14:43 +0000)]
Update docs to add note and examples for tf.count_nonzero with string

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agoMerge pull request #18694 from yongtang/18692-unstack-doc
Martin Wicke [Sun, 22 Apr 2018 06:01:00 +0000 (23:01 -0700)]
Merge pull request #18694 from yongtang/18692-unstack-doc

Update docs for tf.unstack with respect to numpy compatibility

6 years agoMerge pull request #18713 from ManHyuk/fix_typo
Martin Wicke [Sun, 22 Apr 2018 05:54:45 +0000 (22:54 -0700)]
Merge pull request #18713 from ManHyuk/fix_typo

Fix typo

6 years agoMerge pull request #18757 from foo0x29a/master
Martin Wicke [Sun, 22 Apr 2018 05:18:10 +0000 (22:18 -0700)]
Merge pull request #18757 from foo0x29a/master

Fix typo

6 years agoMerge pull request #18168 from imsheridan/fix_adam_math
Martin Wicke [Sun, 22 Apr 2018 05:08:48 +0000 (22:08 -0700)]
Merge pull request #18168 from imsheridan/fix_adam_math

Fix adam optimizer related math equation rendering format

6 years agoupdate $ source spacing
Bryan Heden [Sun, 22 Apr 2018 00:45:42 +0000 (19:45 -0500)]
update $ source spacing

When viewing install_linux, the spacing was off for 'Next Steps' section.

6 years agoChanged calls to the depreacted StringPiece::contains with str_util::StrContains
AG Ramesh [Sat, 21 Apr 2018 22:25:37 +0000 (15:25 -0700)]
Changed calls to the depreacted StringPiece::contains with str_util::StrContains

6 years agoUpdate .gitignore for cmake generated files
Yong Tang [Sat, 21 Apr 2018 18:42:03 +0000 (18:42 +0000)]
Update .gitignore for cmake generated files

After running cmake on Linux with:
```
tensorflow/tools/ci_build/ci_build.sh CMAKE tensorflow/tools/ci_build/builds/cmake.sh
```

the following file is left:
```
ubuntu@ubuntu:~/tensorflow$ git status
On branch master
Your branch is up-to-date with 'origin/master'.
Untracked files:
  (use "git add <file>..." to include in what will be committed)

        api_init_files_list.txt

nothing added to commit but untracked files present (use "git add" to track)
ubuntu@ubuntu:~/tensorflow$
```

This fix updates the .gitignore file so that cmake generated files
is not added with git inadvertently.

Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
6 years agofix typo
foo0x29a [Sat, 21 Apr 2018 16:30:52 +0000 (13:30 -0300)]
fix typo

6 years agoSpecify float32 as float type instead of float64
ADiegoCAlonso [Sat, 21 Apr 2018 09:56:10 +0000 (11:56 +0200)]
Specify float32 as float type instead of float64

6 years agoAdd __init__py
ADiegoCAlonso [Sat, 21 Apr 2018 09:43:51 +0000 (11:43 +0200)]
Add __init__py

6 years agoMerge commit for internal changes
Yifei Feng [Fri, 20 Apr 2018 23:43:26 +0000 (16:43 -0700)]
Merge commit for internal changes

6 years ago[XLA] Remove constant cast in literal util.
Yunxing Dai [Fri, 20 Apr 2018 23:30:02 +0000 (16:30 -0700)]
[XLA] Remove constant cast in literal util.

It's not portable to modify an underlying char array of a c++ string object: (https://stackoverflow.com/questions/5729203/modifying-underlying-char-array-of-a-c-string-object)

RELNOTES: n/a
PiperOrigin-RevId: 193740595

6 years agotf.contrib.data.scan: Support eager execution.
Asim Shankar [Fri, 20 Apr 2018 23:18:29 +0000 (16:18 -0700)]
tf.contrib.data.scan: Support eager execution.

PiperOrigin-RevId: 193739234

6 years agoFix casting in Keras estimator (#18104)
Frédéric Branchaud-Charron [Fri, 20 Apr 2018 23:10:41 +0000 (19:10 -0400)]
Fix casting in Keras estimator (#18104)

6 years agoRemove protected data members from GraphOptimizerStage.
A. Unique TensorFlower [Fri, 20 Apr 2018 23:05:47 +0000 (16:05 -0700)]
Remove protected data members from GraphOptimizerStage.

PiperOrigin-RevId: 193737654

6 years agoSupport legacy clusters
Brennan Saeta [Fri, 20 Apr 2018 22:51:16 +0000 (15:51 -0700)]
Support legacy clusters

PiperOrigin-RevId: 193735742

6 years agoAlways use the local worker name in CreateWorkerSession when not doing ClusterSpec...
Derek Murray [Fri, 20 Apr 2018 22:38:06 +0000 (15:38 -0700)]
Always use the local worker name in CreateWorkerSession when not doing ClusterSpec propagation.

Previously, the master would send a job name and task index in an
otherwise-empty ServerDef, and the worker would unquestioningly use
those to build its worker name. However, this would lead to errors if
the worker had a local name like "/job:worker/replica:1/task:0",
because the ServerDef doesn't support non-zero replica IDs, and so the
local worker would end up an inconsistent view of what its worker name
should be. In particular `WorkerSession::worker_name` would disagree
with the device names added during graph partitioning by the master,
which would lead to runtime failures ("InvalidArgumentError: Invalid
rendezvous key").

PiperOrigin-RevId: 193733855

6 years ago[XLA] Redesign: implement DumpToDirectory for the HloSession.
A. Unique TensorFlower [Fri, 20 Apr 2018 22:32:32 +0000 (15:32 -0700)]
[XLA] Redesign: implement DumpToDirectory for the HloSession.

This is to prepare the migration of tf2xla.

PiperOrigin-RevId: 193733029

6 years ago[XLA] Redesign: add comparator and printer for the XlaOp.
A. Unique TensorFlower [Fri, 20 Apr 2018 22:20:37 +0000 (15:20 -0700)]
[XLA] Redesign: add comparator and printer for the XlaOp.

This is to prepare the migration of tf2xla. There were some codes used ComputationDataHandle::handle() for comparison/printing. Now implement XlaOp's comparator and printer.

PiperOrigin-RevId: 193731437

6 years agoDo not force default layout when there is no need to.
A. Unique TensorFlower [Fri, 20 Apr 2018 22:19:59 +0000 (15:19 -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.

PiperOrigin-RevId: 193731341

6 years agoMove the guts of TFE_Execute into EagerExecute
Akshay Modi [Fri, 20 Apr 2018 21:58:56 +0000 (14:58 -0700)]
Move the guts of TFE_Execute into EagerExecute

PiperOrigin-RevId: 193728072

6 years agointernal
A. Unique TensorFlower [Fri, 20 Apr 2018 21:32:07 +0000 (14:32 -0700)]
internal
END_PUBLIC

BEGIN_PUBLIC
Automated g4 rollback of changelist 193600682

PiperOrigin-RevId: 193723856

6 years agoUpdating Generate Random Tensor to generate tensors whose values are small and do...
A. Unique TensorFlower [Fri, 20 Apr 2018 21:30:48 +0000 (14:30 -0700)]
Updating Generate Random Tensor to generate tensors whose values are small and do not cause overflow for arithmetic operations.

PiperOrigin-RevId: 193723661

6 years agoPin rbe-debian8-tf container tp a newer base image
A. Unique TensorFlower [Fri, 20 Apr 2018 21:25:57 +0000 (14:25 -0700)]
Pin rbe-debian8-tf container tp a newer base image

- Also improve how numpy is installed (not compiling from source) for containers based on other distros than Ubuntu14.04

PiperOrigin-RevId: 193722848

6 years agoAutomated g4 rollback of changelist 193694958
Derek Murray [Fri, 20 Apr 2018 20:58:51 +0000 (13:58 -0700)]
Automated g4 rollback of changelist 193694958

PiperOrigin-RevId: 193718607

6 years agoTrack dependencies between outside_compilation clusters so that control edges can...
A. Unique TensorFlower [Fri, 20 Apr 2018 20:56:55 +0000 (13:56 -0700)]
Track dependencies between outside_compilation clusters so that control edges can be correctly added to sequence compiled computations.

PiperOrigin-RevId: 193718295

6 years agoFix for: Suggest braces around initialization of subobject.
Brian Patton [Fri, 20 Apr 2018 20:54:00 +0000 (13:54 -0700)]
Fix for: Suggest braces around initialization of subobject.

PiperOrigin-RevId: 193717872

6 years agoMerged commit includes the following changes:
A. Unique TensorFlower [Fri, 20 Apr 2018 20:48:37 +0000 (13:48 -0700)]
Merged commit includes the following changes:
193717076  by yifeif:

    Automated g4 rollback of changelist 193713153.

--
193716750  by fchollet:

    Refactor `tf.keras.layers.Embedding` layer to use `embedding_lookup` instead of `gather`. This makes the layer TPU-compatible.

--
193716664  by A. Unique TensorFlower:

    Go: Update generated wrapper functions for TensorFlow ops.

--
193713153  by power:

    Experimental Keras TPU compatibility layer.

--

PiperOrigin-RevId: 193717076

6 years agoMKLDNN: conv2d forward DNN primitive reuse enhancement (#17943)
Guozhong Zhuang [Fri, 20 Apr 2018 20:43:06 +0000 (13:43 -0700)]
MKLDNN: conv2d forward DNN primitive reuse enhancement (#17943)

* Enable conv2d fwd primitive reuse

* coding style change based on suggestions from TF team

* minor code style fix

* refactor conv2d primitive reuse class and enhance key creation utility

* refactor by introducing ConvFwdDimensions structure

* change 'Execute' method to be a template one per PR review suggestion

* Per PR review suggestion, update DnnOp class to declared related  method as abstract ones

* refactor AddAsKey method - template for scalar value and remove Execute()which is not used yet

* rename padding_l/_r/pl/pr to padding_left or padding_right as recommended

* parameter and variable renaming - to make them more explicit

6 years agoUpdate ops-related pbtxt files.
A. Unique TensorFlower [Fri, 20 Apr 2018 20:18:02 +0000 (13:18 -0700)]
Update ops-related pbtxt files.

PiperOrigin-RevId: 193712839

6 years agoAdd a utility to visualize object-based checkpoints
Allen Lavoie [Fri, 20 Apr 2018 19:40:57 +0000 (12:40 -0700)]
Add a utility to visualize object-based checkpoints

Useful for generating a warm fuzzy feeling that everything you think should be saved was saved, and for explaining what object-based checkpointing is. (Also useful on the former front will be a planned "assert that all of this Graph's trainable variables are accessible from object X" function.)

Somewhat hacky since it generates strings rather than using the pydot bindings (and so works without a pydot dependency).

PiperOrigin-RevId: 193708003

6 years agoChanged tf_to_tflite build rule.
A. Unique TensorFlower [Fri, 20 Apr 2018 19:37:39 +0000 (12:37 -0700)]
Changed tf_to_tflite build rule.

PiperOrigin-RevId: 193707628

6 years ago[INTEL MKLDNN]: Upgrade mkldnn version to v13 (#18508)
jinghuangintel [Fri, 20 Apr 2018 19:20:00 +0000 (12:20 -0700)]
[INTEL MKLDNN]: Upgrade mkldnn version to v13 (#18508)

* upgrade mkldnn version to v13

* upgrade mkldnn version to v13 for all platforms

6 years agoAdd native dilated support for conv3d and its gradients in cudnn v>=6.
Yangzihao Wang [Fri, 20 Apr 2018 19:05:11 +0000 (12:05 -0700)]
Add native dilated support for conv3d and its gradients in cudnn v>=6.

PiperOrigin-RevId: 193703316

6 years ago[TF:XLA] Now that the compiler no longer introduces implicit broadcasts, forbid them...
Dimitris Vardoulakis [Fri, 20 Apr 2018 19:02:56 +0000 (12:02 -0700)]
[TF:XLA] Now that the compiler no longer introduces implicit broadcasts, forbid them in the HLO verifier.

PiperOrigin-RevId: 193702874

6 years agoPrefix compat import with underscore in meta_graph_transform.py so that it
Anna R [Fri, 20 Apr 2018 19:01:21 +0000 (12:01 -0700)]
Prefix compat import with underscore in meta_graph_transform.py so that it
doesn't get exported as part of API: https://www.tensorflow.org/versions/r1.8/api_docs/python/tf/contrib/meta_graph_transform/meta_graph_transform

PiperOrigin-RevId: 193702570

6 years ago[XLA] Fix a bug in ToProto: don't add gather attributes twice.
A. Unique TensorFlower [Fri, 20 Apr 2018 18:43:48 +0000 (11:43 -0700)]
[XLA] Fix a bug in ToProto: don't add gather attributes twice.

PiperOrigin-RevId: 193699745

6 years agoMove the guts of TFE_Op into EagerOperation
Akshay Modi [Fri, 20 Apr 2018 18:34:55 +0000 (11:34 -0700)]
Move the guts of TFE_Op into EagerOperation

PiperOrigin-RevId: 193698320

6 years agoAllow creating tensors from numpy arrays, and other various constants - try #2
Akshay Modi [Fri, 20 Apr 2018 18:24:53 +0000 (11:24 -0700)]
Allow creating tensors from numpy arrays, and other various constants - try #2

Allow type-inference from a different input tensor, similar to args_to_matching_eager.

- Update TFE_Py_TensorShapeSlice to take tuples.
- Update int values to allow int/long in py2
END_PUBLIC

BEGIN_PUBLIC
Automated g4 rollback of changelist 192184809

PiperOrigin-RevId: 193696790

6 years agoTemporarily set cudnn Rnn math precision to fp32.
James Qin [Fri, 20 Apr 2018 18:23:29 +0000 (11:23 -0700)]
Temporarily set cudnn Rnn math precision to fp32.

Problem:
When calling cudnnGetRNNLinLayerMatrixParams(), return error CUDNN_STATUS_BAD_PARAM if:

* RNN descriptor set math precision = CUDNN_DATA_FLOAT
* input descriptor dataType = CUDNN_DATA_HALF
* weight descriptor dataType= CUDNN_DATA_HALF

If updating Rnn descriptor math precision to CUDNN_DATA_HALF, then no error.

cudnn 7.1.4 will fix the problem.

PiperOrigin-RevId: 193696566

6 years agoSpeed up safe_strtod and safe_strtof functions by using double-conversion library
akindyakov [Fri, 20 Apr 2018 18:23:15 +0000 (11:23 -0700)]
Speed up safe_strtod and safe_strtof functions by using double-conversion library
Closes #12102.

PiperOrigin-RevId: 193696537

6 years agoUse CreateWorkerSession and DeleteWorkerSession for all distributed sessions.
Derek Murray [Fri, 20 Apr 2018 18:13:16 +0000 (11:13 -0700)]
Use CreateWorkerSession and DeleteWorkerSession for all distributed sessions.

This change adds a phase to the session creation protocol: the master now contacts all workers to register a session handle and create a "WorkerSession" on each worker before it first registers or runs a graph on any worker. Subsequent requests to a worker ensure that the worker has the session handle registered before performing the request, and an AbortedError is raised if the worker has not (e.g. because it restarted after a failure).

As a result, more failure cases are covered by the high-level APIs (tf.estimator, Slim, etc.) that recreate the session on receiving an AbortedError. Previously, there was a possible race condition in which a PS task could restart between variable initialization and the first step, leading to a FailedPreconditionError ("Attempting to use uninitialized value") that would not be handled by the high-level APIs.

PiperOrigin-RevId: 193694958

6 years ago[XLA:CPU] Use Eigen for F64 dot operations
Sanjoy Das [Fri, 20 Apr 2018 18:10:56 +0000 (11:10 -0700)]
[XLA:CPU] Use Eigen for F64 dot operations

PiperOrigin-RevId: 193694613

6 years agoMerge pull request #18663 from imsheridan/fix_community_render
Martin Wicke [Fri, 20 Apr 2018 18:06:46 +0000 (11:06 -0700)]
Merge pull request #18663 from imsheridan/fix_community_render

Fix incorrect format in community/documentation.md

6 years agoDisable constant folding and arithmetic optimizations for functions.
A. Unique TensorFlower [Fri, 20 Apr 2018 17:34:49 +0000 (10:34 -0700)]
Disable constant folding and arithmetic optimizations for functions.

PiperOrigin-RevId: 193688466

6 years agoAdding Shape inference functions to outfeed enqueue ops.
Jacques Pienaar [Fri, 20 Apr 2018 17:32:24 +0000 (10:32 -0700)]
Adding Shape inference functions to outfeed enqueue ops.

PiperOrigin-RevId: 193688099

6 years agoChange the TF record reader to use 16MB buffering by default in order to improve...
A. Unique TensorFlower [Fri, 20 Apr 2018 17:16:03 +0000 (10:16 -0700)]
Change the TF record reader to use 16MB buffering by default in order to improve performance.

PiperOrigin-RevId: 193685521

6 years agoMerge pull request #18714 from yifeif/branch_193610264
Yifei Feng [Fri, 20 Apr 2018 17:08:40 +0000 (10:08 -0700)]
Merge pull request #18714 from yifeif/branch_193610264

Branch 193610264

6 years agoCorrect error in "Adding An Op" docs.
James Keeling [Fri, 20 Apr 2018 16:54:50 +0000 (09:54 -0700)]
Correct error in "Adding An Op" docs.

The macro `REGISTER_KERNEL_BUILDER` always declared a functor specialized on floats, instead of the type actually passed into the macro.

PiperOrigin-RevId: 193682519

6 years agoDon't delete inbound_nodes and outbound_nodes, these no longer exist.
Mark Daoust [Fri, 20 Apr 2018 16:31:08 +0000 (09:31 -0700)]
Don't delete inbound_nodes and outbound_nodes, these no longer exist.

PiperOrigin-RevId: 193679512

6 years agoFix msan error in MapAndBatchDataset.
Saurabh Saxena [Fri, 20 Apr 2018 16:23:52 +0000 (09:23 -0700)]
Fix msan error in MapAndBatchDataset.
While checkpointing tensors in BatchResult.output save only the initialized slice.
If the final batch is short, the entire batch tensor may not be initialized.

PiperOrigin-RevId: 193678679

6 years ago[TF:XLA] Bump open source llvm revision to r330313
Sanjoy Das [Fri, 20 Apr 2018 16:20:36 +0000 (09:20 -0700)]
[TF:XLA] Bump open source llvm revision to r330313

PiperOrigin-RevId: 193678317

6 years agoEliminate the guard around Winograd non-fused convolutions with cudnn7.
A. Unique TensorFlower [Fri, 20 Apr 2018 15:08:01 +0000 (08:08 -0700)]
Eliminate the guard around Winograd non-fused convolutions with cudnn7.

PiperOrigin-RevId: 193669636

6 years agointernal change
A. Unique TensorFlower [Fri, 20 Apr 2018 13:36:56 +0000 (06:36 -0700)]
internal change

PiperOrigin-RevId: 193659701

6 years ago[TF:XLA] Implement ClipByValue.
Peter Hawkins [Fri, 20 Apr 2018 10:29:31 +0000 (03:29 -0700)]
[TF:XLA] Implement ClipByValue.

PiperOrigin-RevId: 193646890

6 years agoUpdate c_api_test.cc
SukHwan Kim [Fri, 20 Apr 2018 09:24:52 +0000 (18:24 +0900)]
Update c_api_test.cc

Typo

6 years ago[TF:XLA] Factor out the handling of while instructions to make HloVerifier::Run shorter.
Dimitris Vardoulakis [Fri, 20 Apr 2018 06:35:04 +0000 (23:35 -0700)]
[TF:XLA] Factor out the handling of while instructions to make HloVerifier::Run shorter.

PiperOrigin-RevId: 193626864

6 years agoAutomated g4 rollback of changelist 193602050
Asim Shankar [Fri, 20 Apr 2018 06:08:53 +0000 (23:08 -0700)]
Automated g4 rollback of changelist 193602050

PiperOrigin-RevId: 193625346

6 years ago[XLA] Rework the local XLA client's Shape class with separate array and tuple shape...
Roy Frostig [Fri, 20 Apr 2018 06:01:07 +0000 (23:01 -0700)]
[XLA] Rework the local XLA client's Shape class with separate array and tuple shape constructors.

PiperOrigin-RevId: 193624591

6 years agoMerge remote-tracking branch 'upstream/master' into ENH/scatter_nd_supports_string
Yan Facai (颜发才) [Fri, 20 Apr 2018 05:38:03 +0000 (13:38 +0800)]
Merge remote-tracking branch 'upstream/master' into ENH/scatter_nd_supports_string

6 years agoTST: byte string for python3
Yan Facai (颜发才) [Fri, 20 Apr 2018 05:33:05 +0000 (13:33 +0800)]
TST: byte string for python3