platform/upstream/tensorflow.git
7 years agoAutomated g4 rollback of changelist 183273334
Eugene Brevdo [Thu, 25 Jan 2018 22:28:41 +0000 (14:28 -0800)]
Automated g4 rollback of changelist 183273334

PiperOrigin-RevId: 183291956

7 years agoDeprecation warning on Variables's += methods
Alexandre Passos [Thu, 25 Jan 2018 22:18:45 +0000 (14:18 -0800)]
Deprecation warning on Variables's += methods

PiperOrigin-RevId: 183290246

7 years agoRemove no longer used param fields from TfLiteResizeBilinerParams
A. Unique TensorFlower [Thu, 25 Jan 2018 21:46:54 +0000 (13:46 -0800)]
Remove no longer used param fields from TfLiteResizeBilinerParams

PiperOrigin-RevId: 183284937

7 years agoVLOG shape inference and annotation return status.
Yao Zhang [Thu, 25 Jan 2018 21:21:24 +0000 (13:21 -0800)]
VLOG shape inference and annotation return status.

PiperOrigin-RevId: 183280222

7 years agoAutomated g4 rollback of changelist 183251689
A. Unique TensorFlower [Thu, 25 Jan 2018 20:24:58 +0000 (12:24 -0800)]
Automated g4 rollback of changelist 183251689

PiperOrigin-RevId: 183273334

7 years agoAdd C0330 bad-continuation check to pylint.
Yifei Feng [Thu, 25 Jan 2018 20:02:36 +0000 (12:02 -0800)]
Add C0330 bad-continuation check to pylint.

PiperOrigin-RevId: 183270896

7 years ago[XLA] Make xla_hlo_profile_test less flaky
Sanjoy Das [Thu, 25 Jan 2018 19:17:08 +0000 (11:17 -0800)]
[XLA] Make xla_hlo_profile_test less flaky

Instead of relying on some oeprations always taking longer than others (and this
appearing in a specific order in the rendered HLO profile), pick them out by
opcode.

PiperOrigin-RevId: 183268593

7 years agoSimplify the template mechanism by specifying templates using multi-line strings...
A. Unique TensorFlower [Thu, 25 Jan 2018 18:36:25 +0000 (10:36 -0800)]
Simplify the template mechanism by specifying templates using multi-line strings instead of functions. This loses the syntax verification on templates, but it avoids the clutter of lint overrides and the duplication of parameter names, so things are more readable.

Addresses #16318

PiperOrigin-RevId: 183260854

7 years agoSurface error if input is not flat list of eager tensors
Akshay Modi [Thu, 25 Jan 2018 17:53:10 +0000 (09:53 -0800)]
Surface error if input is not flat list of eager tensors

PiperOrigin-RevId: 183253112

7 years ago[TF RNN] Ensure dynamic_rnn runs at least one step to propagate dynamic shape info.
Eugene Brevdo [Thu, 25 Jan 2018 17:44:04 +0000 (09:44 -0800)]
[TF RNN] Ensure dynamic_rnn runs at least one step to propagate dynamic shape info.

PiperOrigin-RevId: 183251689

7 years agoAdd functions to encapsulate the logic for checking and setting tensor type.
Nupur Garg [Thu, 25 Jan 2018 17:34:11 +0000 (09:34 -0800)]
Add functions to encapsulate the logic for checking and setting tensor type.

PiperOrigin-RevId: 183250334

7 years agoAdd shard count to select and scatter tests
Yunxing Dai [Thu, 25 Jan 2018 17:32:22 +0000 (09:32 -0800)]
Add shard count to select and scatter tests

RELNOTES: n/a
PiperOrigin-RevId: 183250076

7 years agoAutomated g4 rollback of changelist 183239252
A. Unique TensorFlower [Thu, 25 Jan 2018 16:20:51 +0000 (08:20 -0800)]
Automated g4 rollback of changelist 183239252

PiperOrigin-RevId: 183241034

7 years agoDrop the manually_create field from RnnState.
A. Unique TensorFlower [Thu, 25 Jan 2018 16:05:30 +0000 (08:05 -0800)]
Drop the manually_create field from RnnState.

Initially, I thought that the shape of RNN state arrays could always be
determined by shape propagation. Then I came across some graphs where this
wasn't so easy to infer, so I introduced manually_create thinking of it
as a hack. Today I took another look at dropping that hack, and had a
"D'oh" moment when I realized that the cyclic nature of RNN graphs makes
it impossible to infer the shapes of all arrays by usual propagation.
For example, in a LSTM cell, the input array is concatenated with
a state array, so if we don't already know the shape of that state array,
shape propagation stops there.

Thus, this change removes manually_create by making toco always behave as
if manually_create=true, i.e. early-creating all RNN state arrays with
the shape explicitly specified by the user. The next TODO item here
(see model_flags.proto) is to introduce a generic 'shape' field, so far
the current 'size' field only allows specifying 1-D shapes.

PiperOrigin-RevId: 183239252

7 years agoFixes FunctionLibraryRuntime to be destroyed before ExecutorImpl when NewLocalExecuto...
A. Unique TensorFlower [Thu, 25 Jan 2018 12:48:59 +0000 (04:48 -0800)]
Fixes FunctionLibraryRuntime to be destroyed before ExecutorImpl when NewLocalExecutor returns an error status.

PiperOrigin-RevId: 183220585

7 years agoAdd WarmStartSettings configuration for all Estimators.
A. Unique TensorFlower [Thu, 25 Jan 2018 08:05:16 +0000 (00:05 -0800)]
Add WarmStartSettings configuration for all Estimators.

PiperOrigin-RevId: 183197945

7 years agoFix eager Pooling1D unit test for data_format='channels_first'
Francois Chollet [Thu, 25 Jan 2018 07:36:20 +0000 (23:36 -0800)]
Fix eager Pooling1D unit test for data_format='channels_first'

PiperOrigin-RevId: 183196050

7 years agoAutomated g4 rollback of changelist 183171572
A. Unique TensorFlower [Thu, 25 Jan 2018 06:33:14 +0000 (22:33 -0800)]
Automated g4 rollback of changelist 183171572

PiperOrigin-RevId: 183192221

7 years ago[XLA] Add more tests for ConvertElementType.
Chris Leary [Thu, 25 Jan 2018 04:51:33 +0000 (20:51 -0800)]
[XLA] Add more tests for ConvertElementType.

PiperOrigin-RevId: 183185601

7 years agoMake the graph generation for TFBT deterministic.
A. Unique TensorFlower [Thu, 25 Jan 2018 04:13:14 +0000 (20:13 -0800)]
Make the graph generation for TFBT deterministic.

PiperOrigin-RevId: 183183086

7 years ago[XLA] Fix HloModule clone.
A. Unique TensorFlower [Thu, 25 Jan 2018 03:47:58 +0000 (19:47 -0800)]
[XLA] Fix HloModule clone.

Currently the cloning of an instruction is usually "shallow": the called
computations of the instruction are reused in the clone. This mechanism is
useful when the hlo graph need to be modified in place (e.g. inliner, and some
hlo passes). One exception is the fusion instruction: it's always "deep"
copied, which means the fused computation is copied as well. So when we deep
cloning an HLO module, don't re-copy the fused computation, and do let the
instruction's clone function know where to put the copied fused computation.

PiperOrigin-RevId: 183181206

7 years agoAllow passing candidate sampling seed in to sampled_softmax_loss for testing.
Rui Zhao [Thu, 25 Jan 2018 03:32:10 +0000 (19:32 -0800)]
Allow passing candidate sampling seed in to sampled_softmax_loss for testing.

PiperOrigin-RevId: 183180196

7 years ago[XLA] Allow buffers for CustomCalls to be reused.
Justin Lebar [Thu, 25 Jan 2018 03:01:16 +0000 (19:01 -0800)]
[XLA] Allow buffers for CustomCalls to be reused.

PiperOrigin-RevId: 183177944

7 years agoRefactoring pass.
A. Unique TensorFlower [Thu, 25 Jan 2018 02:23:18 +0000 (18:23 -0800)]
Refactoring pass.
Add a container context class to shorten argument lists and expose context information in a more organized manner.
Clean up names, docs and tests.
Rename submodule to avoid clashing with the new @covert decorator.

PiperOrigin-RevId: 183174971

7 years agoDon't close session in debug_gradients_test.py.
Skye Wanderman-Milne [Thu, 25 Jan 2018 01:56:57 +0000 (17:56 -0800)]
Don't close session in debug_gradients_test.py.

PiperOrigin-RevId: 183171960

7 years agoVISIBILITY_FIX: Add 'auto_correlation' to _allowed_symbols.
Ian Langmore [Thu, 25 Jan 2018 01:56:55 +0000 (17:56 -0800)]
VISIBILITY_FIX:  Add 'auto_correlation' to _allowed_symbols.

PiperOrigin-RevId: 183171955

7 years agoMake batch_sequences_with_states_test.py work with C API enabled.
Skye Wanderman-Milne [Thu, 25 Jan 2018 01:52:19 +0000 (17:52 -0800)]
Make batch_sequences_with_states_test.py work with C API enabled.

PiperOrigin-RevId: 183171572

7 years agoMake resource_variable_ops_test.py work with the C API enabled.
Skye Wanderman-Milne [Thu, 25 Jan 2018 01:49:28 +0000 (17:49 -0800)]
Make resource_variable_ops_test.py work with the C API enabled.

PiperOrigin-RevId: 183171341

7 years agoAdd R3 and R5 tests to select and scatter.
Yunxing Dai [Thu, 25 Jan 2018 01:24:50 +0000 (17:24 -0800)]
Add R3 and R5 tests to select and scatter.

- Added evaluator support so that we can test S&S in arbitrary dimensions.

RELNOTES: n/a
PiperOrigin-RevId: 183168473

7 years agoAdd copy method to EagerVariableStore.
A. Unique TensorFlower [Thu, 25 Jan 2018 01:18:47 +0000 (17:18 -0800)]
Add copy method to EagerVariableStore.

EagerVariableStore.copy creates a new EagerVariableStore instance containing new variables so that they can be modified without affecting the variables in the old store.

PiperOrigin-RevId: 183167776

7 years agoFix comparison bug in remove_trivial_binary
A. Unique TensorFlower [Thu, 25 Jan 2018 01:16:58 +0000 (17:16 -0800)]
Fix comparison bug in remove_trivial_binary

PiperOrigin-RevId: 183167508

7 years agoDedup control dependencies where the node already depends on the source as a non...
A. Unique TensorFlower [Thu, 25 Jan 2018 01:11:42 +0000 (17:11 -0800)]
Dedup control dependencies where the node already depends on the source as a non-control input.

PiperOrigin-RevId: 183166819

7 years agoDefer logging infeed error messages for a short time to see if the main session returns.
Russell Power [Thu, 25 Jan 2018 00:41:50 +0000 (16:41 -0800)]
Defer logging infeed error messages for a short time to see if the main session returns.

PiperOrigin-RevId: 183162866

7 years agoPrepare internal kernels for BroadcastMul and BroadcastAdd
A. Unique TensorFlower [Thu, 25 Jan 2018 00:28:20 +0000 (16:28 -0800)]
Prepare internal kernels for BroadcastMul and BroadcastAdd

PiperOrigin-RevId: 183160907

7 years agoBasic AddN support in toco
A. Unique TensorFlower [Thu, 25 Jan 2018 00:23:09 +0000 (16:23 -0800)]
Basic AddN support in toco

PiperOrigin-RevId: 183160197

7 years ago[TF:XLA:CPU] Fix an issue with explicitly vectorized reductions
Sanjoy Das [Thu, 25 Jan 2018 00:10:37 +0000 (16:10 -0800)]
[TF:XLA:CPU] Fix an issue with explicitly vectorized reductions

The explicitly vectorized reduction kernel assumes that the input and the output
layouts are "same" (not identical which is not possible, but have the same
ordering of the un-reduced dimensions).  This change makes IrEmitter check this
precondition.

PiperOrigin-RevId: 183158249

7 years ago[TF:XLA] Replace most of HloProfilePrinter by a protocol buffer
Sanjoy Das [Thu, 25 Jan 2018 00:10:12 +0000 (16:10 -0800)]
[TF:XLA] Replace most of HloProfilePrinter by a protocol buffer

This change replaces the meat of HloProfilePrinter with a protobuf
HloProfilePrinterData.  The original plan was to serialize HloProfilePrinter
into C++ source code and put that in a .cc file along with the string for the
xla::ProgramShape.  However, since we now directly serialize xla::ProgramShape
into a .o file, for consistency I think we should do the same thing for
HloProfilePrinter (instead of adding yet another output file to tfcompile).

The change itself is fairly simple, it is large mostly due to the mass renaming
I had to do.

PiperOrigin-RevId: 183158192

7 years agoSet export_outs in KMeans' EstimatorSpec.
Yutaka Leon [Wed, 24 Jan 2018 23:45:55 +0000 (15:45 -0800)]
Set export_outs in KMeans' EstimatorSpec.

PiperOrigin-RevId: 183154542

7 years agoMake tensor_util_test.py work with the C API enabled.
Skye Wanderman-Milne [Wed, 24 Jan 2018 23:23:12 +0000 (15:23 -0800)]
Make tensor_util_test.py work with the C API enabled.

Some of the shape inference error messages have changed. We also need
to wrap an InvalidArgumentError as a ValueError for backwards compat.

PiperOrigin-RevId: 183150857

7 years agointernal change
A. Unique TensorFlower [Wed, 24 Jan 2018 23:10:52 +0000 (15:10 -0800)]
internal change

PiperOrigin-RevId: 183148922

7 years agoModify dynamic_rnn to use max_seq_length for its while_loop.
Eugene Brevdo [Wed, 24 Jan 2018 23:10:28 +0000 (15:10 -0800)]
Modify dynamic_rnn to use max_seq_length for its while_loop.

This is especially important with XLA compilation, which will usually have
a maximum time dimension on the input that is (sometimes much larger than)
max(sequence_length).  In these cases, a lot of compute can be saved.

PiperOrigin-RevId: 183148831

7 years agoParameterized docker build now supports a local pip whl file path.
Amit Patankar [Wed, 24 Jan 2018 23:10:19 +0000 (15:10 -0800)]
Parameterized docker build now supports a local pip whl file path.

PiperOrigin-RevId: 183148798

7 years agoWhen a device list API such as TF_DeviceListName() succeeds, set the output status...
A. Unique TensorFlower [Wed, 24 Jan 2018 23:07:21 +0000 (15:07 -0800)]
When a device list API such as TF_DeviceListName() succeeds, set the output status to OK.

PiperOrigin-RevId: 183148319

7 years agoGo: Update generated wrapper functions for TensorFlow ops.
A. Unique TensorFlower [Wed, 24 Jan 2018 22:56:16 +0000 (14:56 -0800)]
Go: Update generated wrapper functions for TensorFlow ops.

PiperOrigin-RevId: 183146357

7 years ago[XLA] Enable F16 for the CPU and GPU backends.
A. Unique TensorFlower [Wed, 24 Jan 2018 22:41:35 +0000 (14:41 -0800)]
[XLA] Enable F16 for the CPU and GPU backends.

Enhance the IR emitter to handle mathematics operations with F16 types.
Add tests for F16 mathematics operations.

PiperOrigin-RevId: 183144087

7 years agoRemove THIRD_PARTY_ from #include guards
Sanjoy Das [Wed, 24 Jan 2018 22:23:02 +0000 (14:23 -0800)]
Remove THIRD_PARTY_ from #include guards

They don't make sense in the open source repository.

PiperOrigin-RevId: 183140889

7 years agoUpdate ops-related pbtxt files.
A. Unique TensorFlower [Wed, 24 Jan 2018 22:18:30 +0000 (14:18 -0800)]
Update ops-related pbtxt files.

PiperOrigin-RevId: 183140179

7 years agoExpose shared_embedding_columns in tf.feature_column.
A. Unique TensorFlower [Wed, 24 Jan 2018 22:17:07 +0000 (14:17 -0800)]
Expose shared_embedding_columns in tf.feature_column.

PiperOrigin-RevId: 183139975

7 years agoAdds loss_reduction argument to canned estimators.
A. Unique TensorFlower [Wed, 24 Jan 2018 22:17:05 +0000 (14:17 -0800)]
Adds loss_reduction argument to canned estimators.

PiperOrigin-RevId: 183139970

7 years ago[Eager mode] Add an Eager-optimized version of the IteratorGetNext op.
Derek Murray [Wed, 24 Jan 2018 22:03:09 +0000 (14:03 -0800)]
[Eager mode] Add an Eager-optimized version of the IteratorGetNext op.

Since in eager mode the kernel execution happens on the calling thread
(which would be blocked anyway), we can invoke the iterator
synchronously, and do not need to perform a context switch to and from
a background thread.

Improves the latency benchmarks in
//third_party/tensorflow/contrib/eager/python:datasets_test by
approximately 7us to 13us per element.

PiperOrigin-RevId: 183137488

7 years agoSilent copies for int32 tensors in TFE.
Alexandre Passos [Wed, 24 Jan 2018 22:02:14 +0000 (14:02 -0800)]
Silent copies for int32 tensors in TFE.

Fixes #16106

PiperOrigin-RevId: 183137298

7 years ago[TF:XLA] Bump open source llvm revision to r323309
Sanjoy Das [Wed, 24 Jan 2018 21:39:02 +0000 (13:39 -0800)]
[TF:XLA] Bump open source llvm revision to r323309

PiperOrigin-RevId: 183133556

7 years agoMake TFLite Pad op have parity with TF Pad op.
Nupur Garg [Wed, 24 Jan 2018 21:21:54 +0000 (13:21 -0800)]
Make TFLite Pad op have parity with TF Pad op.

PiperOrigin-RevId: 183131070

7 years agoPortability fix for parse_testdata.cc
A. Unique TensorFlower [Wed, 24 Jan 2018 21:09:10 +0000 (13:09 -0800)]
Portability fix for parse_testdata.cc

PiperOrigin-RevId: 183129228

7 years agotf_contextlib forgotten oops
Alexandre Passos [Wed, 24 Jan 2018 20:47:00 +0000 (12:47 -0800)]
tf_contextlib forgotten oops

PiperOrigin-RevId: 183126126

7 years agoInitializes the context before setting the device policy
Alexandre Passos [Wed, 24 Jan 2018 20:46:09 +0000 (12:46 -0800)]
Initializes the context before setting the device policy

PiperOrigin-RevId: 183126037

7 years ago[XLA] Fix crash in HloOrdering::ToString().
Justin Lebar [Wed, 24 Jan 2018 20:12:02 +0000 (12:12 -0800)]
[XLA] Fix crash in HloOrdering::ToString().

HloOrdering::ToString() was crashing when given a computation containing
a fusion instruction.
PiperOrigin-RevId: 183121645

7 years agoInternal change.
Anna R [Wed, 24 Jan 2018 19:44:55 +0000 (11:44 -0800)]
Internal change.

PiperOrigin-RevId: 183117665

7 years ago[BatchNorm] Remove CPU implementation
Yunxing Dai [Wed, 24 Jan 2018 19:42:22 +0000 (11:42 -0800)]
[BatchNorm] Remove CPU implementation

We now use batchnorm rewriter (tensorflow/compiler/xla/service/batchnorm_rewriter.h) to expand batch norm into smaller ops. A specific implementation should not be needed anymore (for CPU).

RELNOTES:n/a
PiperOrigin-RevId: 183117252

7 years agoFixed linter errors.
Jianwei Xie [Wed, 24 Jan 2018 19:31:06 +0000 (11:31 -0800)]
Fixed linter errors.

PiperOrigin-RevId: 183115307

7 years agofix typos in schema.fbs comments.
A. Unique TensorFlower [Wed, 24 Jan 2018 19:29:08 +0000 (11:29 -0800)]
fix typos in schema.fbs comments.

PiperOrigin-RevId: 183114994

7 years agoIntroduce experimental quantize training and eval apis.
Suharsh Sivakumar [Wed, 24 Jan 2018 18:57:28 +0000 (10:57 -0800)]
Introduce experimental quantize training and eval apis.
These APIs will include options that may have undefined behavior, still being researched. The non experimental APIs will have better hardware support and will be more suitable for users aiming to rewrite graphs to provide to backends (e.g. TFLite).

PiperOrigin-RevId: 183109634

7 years agoFix update_api_def.sh script to output ApiDefs instead of ApiDef proto.
Anna R [Wed, 24 Jan 2018 18:55:55 +0000 (10:55 -0800)]
Fix update_api_def.sh script to output ApiDefs instead of ApiDef proto.

PiperOrigin-RevId: 183109303

7 years agoInternal Change
A. Unique TensorFlower [Wed, 24 Jan 2018 18:35:22 +0000 (10:35 -0800)]
Internal Change

PiperOrigin-RevId: 183105591

7 years agoTFE: Add graph benchmark for linear regression, for comparison with our eager executi...
Akshay Agrawal [Wed, 24 Jan 2018 18:34:07 +0000 (10:34 -0800)]
TFE: Add graph benchmark for linear regression, for comparison with our eager execution benchmark.

PiperOrigin-RevId: 183105403

7 years agoMerge changes from github.
Jianwei Xie [Wed, 24 Jan 2018 18:02:35 +0000 (10:02 -0800)]
Merge changes from github.

PiperOrigin-RevId: 183100142

7 years ago* BUGFIX: See code associated with scale_identity_multiplier
Ian Langmore [Wed, 24 Jan 2018 17:49:47 +0000 (09:49 -0800)]
* BUGFIX:  See code associated with scale_identity_multiplier
* BUGFIX:  See code associated with 'weight' inside sample_n
* Use parameterization `temperature` rather than `mix_scale`.
* Simplify documentation and link to arXiv paper for details
* Document that we allow `temperature` to have any shape broadcastable with `mix_loc`.  This is a mute point to some degree since we require K = 2 now.
* Add some tests

PiperOrigin-RevId: 183098275

7 years agoRename internal field.
Jianwei Xie [Wed, 24 Jan 2018 17:15:32 +0000 (09:15 -0800)]
Rename internal field.

PiperOrigin-RevId: 183093901

7 years agoAllow setting per-thread device copying policies in TFE.
Alexandre Passos [Wed, 24 Jan 2018 17:10:53 +0000 (09:10 -0800)]
Allow setting per-thread device copying policies in TFE.

PiperOrigin-RevId: 183093407

7 years agoDon't set unused Operation private fields if the C API is enabled.
Skye Wanderman-Milne [Wed, 24 Jan 2018 17:01:25 +0000 (09:01 -0800)]
Don't set unused Operation private fields if the C API is enabled.

This helps catch invalid uses of the private fields with the C API
enabled, and refactors Operation.__init__ in preparation for enabling
the C API and eventually getting rid of the _USE_C_API toggle.

PiperOrigin-RevId: 183092099

7 years agoAdds sharding in head_test to prevent timeouts.
A. Unique TensorFlower [Wed, 24 Jan 2018 16:35:13 +0000 (08:35 -0800)]
Adds sharding in head_test to prevent timeouts.

PiperOrigin-RevId: 183089222

7 years agoAutomated g4 rollback of changelist 183076718
Jacques Pienaar [Wed, 24 Jan 2018 16:00:10 +0000 (08:00 -0800)]
Automated g4 rollback of changelist 183076718

PiperOrigin-RevId: 183085489

7 years agoRemove unused ResizeBilinear options. Use a tensor to read new sizes.
A. Unique TensorFlower [Wed, 24 Jan 2018 15:59:17 +0000 (07:59 -0800)]
Remove unused ResizeBilinear options. Use a tensor to read new sizes.

PiperOrigin-RevId: 183085403

7 years ago[TPU] Raise error for unimplemented TPU op.
Jacques Pienaar [Wed, 24 Jan 2018 14:21:09 +0000 (06:21 -0800)]
[TPU] Raise error for unimplemented TPU op.

Raise an error as soon as detected that an unsupported TPU op is added. Update the list of unimplemented ops.

PiperOrigin-RevId: 183076718

7 years agoUse RunConfig.master to set up the Session for running prediction in Estimator.predict()
A. Unique TensorFlower [Wed, 24 Jan 2018 09:17:31 +0000 (01:17 -0800)]
Use RunConfig.master to set up the Session for running prediction in Estimator.predict()

PiperOrigin-RevId: 183053484

7 years agoAdd Real NVP and NICE bijectors and tests.
A. Unique TensorFlower [Wed, 24 Jan 2018 07:17:06 +0000 (23:17 -0800)]
Add Real NVP and NICE bijectors and tests.

PiperOrigin-RevId: 183044638

7 years agoMake an API for non-slot variables. This new API is safer than the
A. Unique TensorFlower [Wed, 24 Jan 2018 05:51:06 +0000 (21:51 -0800)]
Make an API for non-slot variables. This new API is safer than the
existing AdamOptimizer code when switching between multiple graphs
or multiple threads.

PiperOrigin-RevId: 183039296

7 years ago[tf.data] Fixes flaky timeouts of stats_dataset_ops_test.
Shivani Agrawal [Wed, 24 Jan 2018 04:09:58 +0000 (20:09 -0800)]
[tf.data] Fixes flaky timeouts of stats_dataset_ops_test.

PiperOrigin-RevId: 183033139

7 years agoAdd nested support for `assertAllClose()` so `a` and `b` can be arbitrarily nested of
A. Unique TensorFlower [Wed, 24 Jan 2018 02:49:09 +0000 (18:49 -0800)]
Add nested support for `assertAllClose()` so `a` and `b` can be arbitrarily nested of
a numpy array, tuple, list, namedtuple or anything can be converted to a numpy array.

PiperOrigin-RevId: 183027752

7 years agoGo: Update generated wrapper functions for TensorFlow ops.
A. Unique TensorFlower [Wed, 24 Jan 2018 02:46:01 +0000 (18:46 -0800)]
Go: Update generated wrapper functions for TensorFlow ops.

PiperOrigin-RevId: 183027491

7 years agoUpdate ops-related pbtxt files.
A. Unique TensorFlower [Wed, 24 Jan 2018 02:18:15 +0000 (18:18 -0800)]
Update ops-related pbtxt files.

PiperOrigin-RevId: 183024949

7 years agoSupport StridedSlice in TFLite for 1D-4D tensors.
A. Unique TensorFlower [Wed, 24 Jan 2018 01:35:53 +0000 (17:35 -0800)]
Support StridedSlice in TFLite for 1D-4D tensors.

PiperOrigin-RevId: 183020501

7 years ago[TPU] Add extra type checking in Python around the infeed operation, since it is...
Peter Hawkins [Wed, 24 Jan 2018 01:32:39 +0000 (17:32 -0800)]
[TPU] Add extra type checking in Python around the infeed operation, since it is a common mistake to make.

PiperOrigin-RevId: 183020077

7 years agoMark TensorArrayV2 ops and deprecated io_ops deprecated.
Jacques Pienaar [Wed, 24 Jan 2018 01:26:50 +0000 (17:26 -0800)]
Mark TensorArrayV2 ops and deprecated io_ops deprecated.

PiperOrigin-RevId: 183019408

7 years ago[XLA:python] Pull static method off of ComputationBuilder class.
Chris Leary [Wed, 24 Jan 2018 01:11:30 +0000 (17:11 -0800)]
[XLA:python] Pull static method off of ComputationBuilder class.

PiperOrigin-RevId: 183017700

7 years agoAdds a mechanism for whitelisting ops for use in map_fns using a WHITELIST_OP_FOR_DAT...
Saurabh Saxena [Wed, 24 Jan 2018 00:50:47 +0000 (16:50 -0800)]
Adds a mechanism for whitelisting ops for use in map_fns using a WHITELIST_OP_FOR_DATASET_FUNCTIONS macro.
Changes criteria for whitelisting stateful dataset ops. We used to check for the presence of the output_shapes attr. Now we just match the name and the output type.

PiperOrigin-RevId: 183014929

7 years agoEnable and fix some bfloat16 tests.
A. Unique TensorFlower [Wed, 24 Jan 2018 00:38:55 +0000 (16:38 -0800)]
Enable and fix some bfloat16 tests.

PiperOrigin-RevId: 183013346

7 years agoAdd unidirectional sequence LSTM to TFLite Ops.
A. Unique TensorFlower [Wed, 24 Jan 2018 00:22:43 +0000 (16:22 -0800)]
Add unidirectional sequence LSTM to TFLite Ops.

PiperOrigin-RevId: 183010993

7 years agoMake the toco python wrapper lazy loaded to avoid dep error.
Andrew Selle [Wed, 24 Jan 2018 00:03:41 +0000 (16:03 -0800)]
Make the toco python wrapper lazy loaded to avoid dep error.

PiperOrigin-RevId: 183007943

7 years agoMake namedtuples with identical name and field names to be considered as the same...
Frank Perbet [Tue, 23 Jan 2018 23:59:09 +0000 (15:59 -0800)]
Make namedtuples with identical name and field names to be considered as the same shallow structure.

PiperOrigin-RevId: 183007218

7 years agoAdd a base transformer class with useful utilities.
A. Unique TensorFlower [Tue, 23 Jan 2018 23:41:33 +0000 (15:41 -0800)]
Add a base transformer class with useful utilities.
Intercept errors in the base class and include additional context information.
Use the base transformer class with type_info.py.

PiperOrigin-RevId: 183004622

7 years ago[TF:XLA:CPU] Alias tensorflow::gtl to gtl for brevity
Sanjoy Das [Tue, 23 Jan 2018 23:20:43 +0000 (15:20 -0800)]
[TF:XLA:CPU] Alias tensorflow::gtl to gtl for brevity

PiperOrigin-RevId: 183001493

7 years agoAdds regularization_losses in remaining heads.
A. Unique TensorFlower [Tue, 23 Jan 2018 23:11:22 +0000 (15:11 -0800)]
Adds regularization_losses in remaining heads.

PiperOrigin-RevId: 183000075

7 years agoSwitch models/ to use new test-specification format
A. Unique TensorFlower [Tue, 23 Jan 2018 23:08:26 +0000 (15:08 -0800)]
Switch models/ to use new test-specification format

PiperOrigin-RevId: 182999650

7 years agoLabeledTensor: don't test Tensor.shape object, test value instead.
Skye Wanderman-Milne [Tue, 23 Jan 2018 23:02:46 +0000 (15:02 -0800)]
LabeledTensor: don't test Tensor.shape object, test value instead.

With the C API enabled, a new TensorShape object may be generated
every Tensor.get_shape call.

PiperOrigin-RevId: 182998609

7 years agoTweaks to PR 16101 which adds stream selection support for `tf.contrib.ffmpeg.decode_...
Yong Tang [Tue, 23 Jan 2018 23:01:14 +0000 (15:01 -0800)]
Tweaks to PR 16101 which adds stream selection support for `tf.contrib.ffmpeg.decode_audio`.

PiperOrigin-RevId: 182998275

7 years agoMake bijectors/reshape_test.py work with C API enabled.
Skye Wanderman-Milne [Tue, 23 Jan 2018 22:54:22 +0000 (14:54 -0800)]
Make bijectors/reshape_test.py work with C API enabled.

Some of the error messages change with the C API enabled due to slight
differences in shape inference (in this case, the C API catches shape
errors sooner). This change refactors some of the tests to expect
different error messages in different cases.

PiperOrigin-RevId: 182997183

7 years agoTurn private Operation and Tensor fields into properties.
Skye Wanderman-Milne [Tue, 23 Jan 2018 22:31:15 +0000 (14:31 -0800)]
Turn private Operation and Tensor fields into properties.

Unfortunately there are many uses of these private fields. Turning
them into properties that call public API methods will allow enabling
the C API before fixing all such uses.

PiperOrigin-RevId: 182993539

7 years agoReplace the use of ConsumeValueOrDie with TF_ASSERT_OK_AND_ASSIGN.
A. Unique TensorFlower [Tue, 23 Jan 2018 22:25:36 +0000 (14:25 -0800)]
Replace the use of ConsumeValueOrDie with TF_ASSERT_OK_AND_ASSIGN.

PiperOrigin-RevId: 182992618

7 years agoUpdating the update_version script to act appropriately for rc0 when it comes
Amit Patankar [Tue, 23 Jan 2018 22:24:53 +0000 (14:24 -0800)]
Updating the update_version script to act appropriately for rc0 when it comes
to the install sources tested configurations table along with a warning. Using perl rather than sed as it supports regex look behind.

PiperOrigin-RevId: 182992467