platform/upstream/tensorflow.git
6 years agoFixes for PR comments and build failures
Sami Kama [Fri, 2 Feb 2018 22:36:07 +0000 (14:36 -0800)]
Fixes for PR comments and build failures

6 years ago[update]
Jie [Fri, 2 Feb 2018 22:08:35 +0000 (14:08 -0800)]
[update]
  removing commented code & unused code

6 years agoFix py3 conversion
Sami Kama [Fri, 2 Feb 2018 18:13:41 +0000 (10:13 -0800)]
Fix py3 conversion

6 years ago[update]
Jie [Thu, 1 Feb 2018 23:15:57 +0000 (15:15 -0800)]
[update]
  addressing comment on variable names

6 years ago[update]
Jie [Thu, 1 Feb 2018 22:48:47 +0000 (14:48 -0800)]
[update]
  addressing comment issues

6 years agoFix ldscript for exporting PyInit_* functions for python3
Sami Kama [Thu, 1 Feb 2018 20:21:05 +0000 (12:21 -0800)]
Fix ldscript for exporting PyInit_* functions for python3

6 years agoAdd/fix copyright and fix format for the test script and other files.
gracehoney [Thu, 1 Feb 2018 06:35:45 +0000 (22:35 -0800)]
Add/fix copyright and fix format for the test script and other files.

6 years agoFix missing defines in compilation flags that caused missing ops and add a simple...
Sami Kama [Thu, 1 Feb 2018 02:49:26 +0000 (18:49 -0800)]
Fix missing defines in compilation flags that caused missing ops and add a simple test to validate functionality

6 years agoChange VLOGs to increasing verbosity mode
Sami Kama [Wed, 31 Jan 2018 19:07:25 +0000 (11:07 -0800)]
Change VLOGs to increasing verbosity mode

6 years agoAdd missing tf_copts() to Build file so #if GOOGLE_CUDA directives work correctly
Sami Kama [Tue, 30 Jan 2018 23:24:11 +0000 (15:24 -0800)]
Add missing tf_copts() to Build file so #if GOOGLE_CUDA directives work correctly

6 years ago[COMMENT]
Jie [Tue, 30 Jan 2018 23:00:38 +0000 (15:00 -0800)]
[COMMENT]
  trt_engine_op compute
    comment on the scope of pointer arrays for i/o binding
    tensorflow/contrib/tensorrt/kernels/trt_engine_op.cc

6 years agoChanges for more PR comments
Sami Kama [Tue, 30 Jan 2018 22:42:53 +0000 (14:42 -0800)]
Changes for more PR comments

6 years agoMerge branch 'tensorrt' of github.com:samikama/tensorflow into tensorrt
Sami Kama [Tue, 30 Jan 2018 22:14:33 +0000 (14:14 -0800)]
Merge branch 'tensorrt' of github.com:samikama/tensorflow into tensorrt

6 years agoFix dependency problem
Sami Kama [Tue, 30 Jan 2018 22:12:54 +0000 (14:12 -0800)]
Fix dependency problem

6 years agoRun clang-format on c++ files
gracehoney [Tue, 30 Jan 2018 21:37:12 +0000 (13:37 -0800)]
Run clang-format on c++ files

6 years agoRemove unused targets
gracehoney [Tue, 30 Jan 2018 21:36:21 +0000 (13:36 -0800)]
Remove unused targets

6 years agoFix the build dependencies and formatting of the code, and make sure
gracehoney [Tue, 30 Jan 2018 20:00:23 +0000 (12:00 -0800)]
Fix the build dependencies and formatting of the code, and make sure
they follow the style conventions.

6 years agoRemove commented lines in configure.py
gracehoney [Tue, 30 Jan 2018 17:05:40 +0000 (09:05 -0800)]
Remove commented lines in configure.py

6 years agoFix GOOGLE_TENSORRT
gracehoney [Tue, 30 Jan 2018 16:36:09 +0000 (08:36 -0800)]
Fix GOOGLE_TENSORRT

6 years agoAdd TODO message
Sami Kama [Tue, 30 Jan 2018 01:34:07 +0000 (17:34 -0800)]
Add TODO message

6 years agoUpdate Readme and add GOOGLE_CUDA conditionals back
Sami Kama [Tue, 30 Jan 2018 01:18:37 +0000 (17:18 -0800)]
Update Readme and add GOOGLE_CUDA conditionals back

6 years agoRemove debug log and use vlog instead
Sami Kama [Tue, 30 Jan 2018 00:10:13 +0000 (16:10 -0800)]
Remove debug log and use vlog instead

6 years agoResolve merge conflict
gracehoney [Mon, 29 Jan 2018 23:53:07 +0000 (15:53 -0800)]
Resolve merge conflict

6 years agoFix merge problems
Sami Kama [Mon, 29 Jan 2018 23:10:57 +0000 (15:10 -0800)]
Fix merge problems

6 years agoFix compilation issues with new build files
Sami Kama [Mon, 29 Jan 2018 23:08:26 +0000 (15:08 -0800)]
Fix compilation issues with new build files

6 years agoFix build (part1):
gracehoney [Mon, 29 Jan 2018 22:49:13 +0000 (14:49 -0800)]
Fix build (part1):
1. Changed includes of "NvInfer.h" to "tensorrt/include/NvInfer.h"
2. Remove build target "tensorrt_ops.so" (src file doesn't exist and the
   target is not used anywhere)
3. Add missing '#if GOOGLE_TENSORRT's
4. Use tf_cuda_library instead of cc_library for some targets to get the
   tf_copts naturally.
5. Revert the changes that was accidentally made (by merging with
   upstream head) from configure.py
6. Replace exception with LOG(FATAL) in
   tensorflow/contrib/tensorrt/convert/convert_nodes.cc as exception is
   not supported.
7. Revert the reinterprete_cast change in
   tensorflow/contrib/tensorrt/kernels/trt_engine_op.cc
8. Fix minor formatting and naming issues according to the style guide.

6 years ago[update]
Jie [Mon, 29 Jan 2018 22:23:54 +0000 (14:23 -0800)]
[update]
  clang-format & cpplint done.

  TODO: compile and test after build fix.

6 years ago[update]
Jie [Mon, 29 Jan 2018 21:26:27 +0000 (13:26 -0800)]
[update]
  addressing:
  i. some naming conventions
  ii. add macro to guard cpp files

  TODO:
    cpplint

6 years agoFix compilation, there are still linking issues which will be fixed in a followup...
gracehoney [Mon, 29 Jan 2018 19:15:32 +0000 (11:15 -0800)]
Fix compilation, there are still linking issues which will be fixed in a followup commit.

6 years ago[UPDATE]
Jie [Mon, 29 Jan 2018 17:56:59 +0000 (09:56 -0800)]
[UPDATE]
  addressing code review comments.

  TODO:
    MACRO GOOGLE_TENSORRT to guard c++ files.
    build/test -> current code works on local repo before the master merge.

6 years agoMerge branch 'master' into tensorrt
gracehoney [Fri, 26 Jan 2018 17:49:23 +0000 (09:49 -0800)]
Merge branch 'master' into tensorrt

6 years agoDecoding contents of BMP file on big endian (#16145)
namrata-ibm [Fri, 26 Jan 2018 16:40:04 +0000 (22:10 +0530)]
Decoding contents of BMP file on big endian (#16145)

* Decoding contents of BMP file on big endian

* Updated as per review comments

* Update decode_bmp_op.cc

Corrected function name

6 years agoUpdate contrib/HVX readme (#16131)
Santiago Castro [Fri, 26 Jan 2018 16:39:07 +0000 (13:39 -0300)]
Update contrib/HVX readme (#16131)

* Update contrib/HVX readme

* Fix PR comments

6 years agoFixing the url for the pip3. (#16447)
Keiji Ariyama [Fri, 26 Jan 2018 16:33:29 +0000 (01:33 +0900)]
Fixing the url for the pip3. (#16447)

6 years agoMerge pull request #16457 from xiejw/branch_183374082
xiejw [Fri, 26 Jan 2018 16:30:33 +0000 (08:30 -0800)]
Merge pull request #16457 from xiejw/branch_183374082

Branch 183374082

6 years agosolve push conflict
Jianwei Xie [Fri, 26 Jan 2018 15:39:13 +0000 (07:39 -0800)]
solve push conflict

6 years agoRoll CL 179861781 forward with fix: Wrappers for CUDA 9 warp-synchronous intrinsics.
A. Unique TensorFlower [Fri, 26 Jan 2018 13:15:18 +0000 (05:15 -0800)]
Roll CL 179861781 forward with fix: Wrappers for CUDA 9 warp-synchronous intrinsics.

PiperOrigin-RevId: 183374082

6 years agoRemove dead code
A. Unique TensorFlower [Fri, 26 Jan 2018 13:14:47 +0000 (05:14 -0800)]
Remove dead code

PiperOrigin-RevId: 183374040

6 years agoFix override annotations
A. Unique TensorFlower [Fri, 26 Jan 2018 11:50:58 +0000 (03:50 -0800)]
Fix override annotations

PiperOrigin-RevId: 183367326

6 years agoFix checkpoint_utils.init_from_checkpoint() to be deterministic.
A. Unique TensorFlower [Fri, 26 Jan 2018 09:17:13 +0000 (01:17 -0800)]
Fix checkpoint_utils.init_from_checkpoint() to be deterministic.

PiperOrigin-RevId: 183354193

6 years agoSet up TensorRT configurations for external use, and add a test.
Guangda Lai [Fri, 26 Jan 2018 07:59:19 +0000 (23:59 -0800)]
Set up TensorRT configurations for external use, and add a test.

PiperOrigin-RevId: 183347199

6 years ago Adds R1 test for ReduceWindow.
Tayo Oguntebi [Fri, 26 Jan 2018 07:37:20 +0000 (23:37 -0800)]
  Adds R1 test for ReduceWindow.

PiperOrigin-RevId: 183345779

6 years agoGo: Update generated wrapper functions for TensorFlow ops.
A. Unique TensorFlower [Fri, 26 Jan 2018 06:45:59 +0000 (22:45 -0800)]
Go: Update generated wrapper functions for TensorFlow ops.

PiperOrigin-RevId: 183342483

6 years agoFor windows cmake build turn on CMAKE_SUPPRESS_REGENERATION to avoid flaky
Gunhan Gulsoy [Fri, 26 Jan 2018 06:32:15 +0000 (22:32 -0800)]
For windows cmake build turn on CMAKE_SUPPRESS_REGENERATION to avoid flaky
build failures.

PiperOrigin-RevId: 183341561

6 years agoClarified documentation on resize_images.align_corners parameter.
A. Unique TensorFlower [Fri, 26 Jan 2018 05:57:50 +0000 (21:57 -0800)]
Clarified documentation on resize_images.align_corners parameter.

PiperOrigin-RevId: 183339087

6 years agoHotfix/fix android example for focus mode continuous picture - #15487 (#15489)
Armando Fandango [Fri, 26 Jan 2018 05:31:56 +0000 (00:31 -0500)]
Hotfix/fix android example for focus mode continuous picture - #15487 (#15489)

* fixed #15487

* fixed #15487

* simplify and tweak formatting to TF style

6 years agoUpdating error handling in normalize_tuple (#15822)
Luke Schaefer [Fri, 26 Jan 2018 05:31:39 +0000 (00:31 -0500)]
Updating error handling in normalize_tuple (#15822)

On line 83 we test to see if single_value is an int (or able to be cast to an int).

ValueError is fired if `int()` is called with an input like 'asdf' - this is caught and gives a helpful error, using the 'name' param to provide more context.

However, when given an other than a string or int, this is *not* caught - making error messages much more esoteric than the helpful one written out here.

For example, before, I was getting an error:
> line 83, in normalize_tuple
>   int(single_value)
>  TypeError: int() argument must be a string or a number, not 'tuple'

Now I get the more useful:
> ValueError: The `kernel_size` argument must be a tuple of 2 integers.
>  Received: ((0, 3), 50) including element (0, 3) of type <type 'tuple'>

6 years agoFix docstrings in `scan` (#16432)
Taehoon Lee [Fri, 26 Jan 2018 05:30:55 +0000 (14:30 +0900)]
Fix docstrings in `scan` (#16432)

6 years agoIncrease tolerance on TFGAN losses test. fixes #16238 (#16435)
Joel Shor [Fri, 26 Jan 2018 05:30:45 +0000 (21:30 -0800)]
Increase tolerance on TFGAN losses test. fixes #16238 (#16435)

6 years agoLog more info about the ill-formed node in ComputeTransitiveFanin.
Max Galkin [Fri, 26 Jan 2018 04:27:11 +0000 (20:27 -0800)]
Log more info about the ill-formed node in ComputeTransitiveFanin.

PiperOrigin-RevId: 183333452

6 years agointernal change
A. Unique TensorFlower [Fri, 26 Jan 2018 04:26:35 +0000 (20:26 -0800)]
internal change

PiperOrigin-RevId: 183333411

6 years ago[XLA] Add source mapping utility translation unit, use it in the local client.
Chris Leary [Fri, 26 Jan 2018 03:50:17 +0000 (19:50 -0800)]
[XLA] Add source mapping utility translation unit, use it in the local client.

PiperOrigin-RevId: 183331075

6 years ago [tpu:profiler] Add infeed enqueue operation data to tf_op_stats.proto.
A. Unique TensorFlower [Fri, 26 Jan 2018 03:10:13 +0000 (19:10 -0800)]
  [tpu:profiler] Add infeed enqueue operation data to tf_op_stats.proto.

PiperOrigin-RevId: 183328456

6 years agoUpdate tf.keras to the Keras 2.1.3 API.
Francois Chollet [Fri, 26 Jan 2018 03:05:28 +0000 (19:05 -0800)]
Update tf.keras to the Keras 2.1.3 API.

PiperOrigin-RevId: 183328052

6 years agoIncluding common.h with NEON_2_SSE.h (#15743)
Stanislav Levental [Fri, 26 Jan 2018 02:52:59 +0000 (18:52 -0800)]
Including common.h with NEON_2_SSE.h (#15743)

* Including common.h with NEON_2_SSE.h

Including common.h to make sure that USE_NEON is defined in case of NEON_2_SSE.h is used; otherwise USE_NEON will not be propagated to this file and `portable_tensor_utils.h` will be used

* Removing spaces

* Using neon is USE_NEON is defined

* Using common.h to set USE_NEON if applicable

* Adding required dependencies

Since tests are more likely compiled on sse4-enabled machine - if they are run with common.h - it will enable sse4, there are two ways - just enable it and let tests using sse4, or add extra flag check which would require USE_SSE4 flag to be defined. I've added all required dependencies for tests to pass, so it could work.

* Rearranging dependencies according to linter

* Fixing sanity check

* trying to rerun, could be cached

* Triggering build

* triggering build - looks like it stuck

no status for more than 24h

* including x86_64 since NEON_2_SSE4 convertion will allow neon to run on x86 with SSE3,4

* Fixing name

* using darwing for MacOS

* Fixing header inclusion rules

* Fix syntax

* Depend on includes

We depend on both types and compatibility. That's in the "types" rule.

6 years agoFix typo (#16425)
ManHyuk [Fri, 26 Jan 2018 02:52:18 +0000 (11:52 +0900)]
Fix typo (#16425)

* fix typos

* fix typos

6 years agoimport contextmanager in side_effect_guards.py (#16426)
cclauss [Fri, 26 Jan 2018 02:51:55 +0000 (03:51 +0100)]
import contextmanager in side_effect_guards.py (#16426)

6 years agoSet size of test //third_party/tensorflow/python/data/kernel_tests:dataset_from_gener...
A. Unique TensorFlower [Fri, 26 Jan 2018 02:46:51 +0000 (18:46 -0800)]
Set size of test //third_party/tensorflow/python/data/kernel_tests:dataset_from_generator_op_test to medium

It sometimes takes longer than a minute, and thus gets flaky timeouts.

PiperOrigin-RevId: 183326334

6 years agoAllow passing dummy/custom minmax information on a per-array basis,
A. Unique TensorFlower [Fri, 26 Jan 2018 02:42:44 +0000 (18:42 -0800)]
Allow passing dummy/custom minmax information on a per-array basis,
unlike the existing --default_ranges_{min,max} flags which only allowed
to set a single global value for all arrays.

This takes the form of a new embedded message in ModelFlags, which is
its own message so that it can be serialized separately. The command-line
interface is --arrays_extra_info_file=some_proto.pbtxt, i.e. we don't
try to make a command-line-flags-only interface, we mandate putting the info
in a file. The rationale is that users may want to specify custom minmax
for hundreds of arrays, so it would be cumbersome to have that all in a
command line.

This should be considered an experimental feature, in the sense that
in properly quantized models, minmax information is already embedded
in the graph (e.g. in FakeQuant nodes). This is an extension of the
existing --default_ranges_{min,max} feature which had turned out to be
too restrictive for many users.

PiperOrigin-RevId: 183326000

6 years agoAdd input_shape to seq2seq helpers.
Adam Roberts [Fri, 26 Jan 2018 01:58:48 +0000 (17:58 -0800)]
Add input_shape to seq2seq helpers.

PiperOrigin-RevId: 183321394

6 years agoReplace instances of `control_dependencies(None)` with `init_scope` when
Akshay Agrawal [Fri, 26 Jan 2018 01:50:40 +0000 (17:50 -0800)]
Replace instances of `control_dependencies(None)` with `init_scope` when
initializing variables.

Today, when variables are constructed (Resource and otherwise), we lift certain
operations, including the VarHandleOp and initialization ops, out of all
control flow contexts; the mechanism for doing so is entering the context
manager returned by `control_dependencies(None)`. This change replaces various
instances of this mechanism with `init_scope`, which clears control dependencies, lifts ops out of function-building graphs, and pauses the gradient tape. As a result,
variables that are created inside graph functions will be automatically hoisted
into an outer context.

PiperOrigin-RevId: 183320576

6 years agoMove flatbuffer verifier to a separate lib
A. Unique TensorFlower [Fri, 26 Jan 2018 01:32:12 +0000 (17:32 -0800)]
Move flatbuffer verifier to a separate lib

PiperOrigin-RevId: 183318384

6 years agoAdd an option to input a GraphDef.
Yao Zhang [Fri, 26 Jan 2018 01:27:42 +0000 (17:27 -0800)]
Add an option to input a GraphDef.

PiperOrigin-RevId: 183317862

6 years agoMake kernel_tests/scalar_test.py work with the C API enabled.
Skye Wanderman-Milne [Fri, 26 Jan 2018 01:20:07 +0000 (17:20 -0800)]
Make kernel_tests/scalar_test.py work with the C API enabled.

This also moves the set_producer_version function from a specific test
file to test_util.py, since it's needed in two test files now.

PiperOrigin-RevId: 183316990

6 years ago[XLA] Disable half_test_cpu as it is flaky
A. Unique TensorFlower [Fri, 26 Jan 2018 01:08:50 +0000 (17:08 -0800)]
[XLA] Disable half_test_cpu as it is flaky

PiperOrigin-RevId: 183315762

6 years agoRecord requested cpu cores in OpPerformance.
Yuefeng Zhou [Fri, 26 Jan 2018 00:50:24 +0000 (16:50 -0800)]
Record requested cpu cores in OpPerformance.

PiperOrigin-RevId: 183313321

6 years agoAutomated g4 rollback of changelist 183296506
Justin Lebar [Fri, 26 Jan 2018 00:45:52 +0000 (16:45 -0800)]
Automated g4 rollback of changelist 183296506

PiperOrigin-RevId: 183312680

6 years agoDeleted unused data fields
Benoit Steiner [Fri, 26 Jan 2018 00:45:04 +0000 (16:45 -0800)]
Deleted unused data fields

PiperOrigin-RevId: 183312596

6 years agoimport tensorflow as tf (#16318)
cclauss [Fri, 26 Jan 2018 00:31:41 +0000 (01:31 +0100)]
import tensorflow as tf (#16318)

* import tensorflow as tf

* import tensorflow as tf

* from contextlib import contextmanager

* remove the last remaining change to py2tf

6 years agoMake batch_sequences_with_states_test.py work with C API enabled, take 2.
Skye Wanderman-Milne [Fri, 26 Jan 2018 00:15:50 +0000 (16:15 -0800)]
Make batch_sequences_with_states_test.py work with C API enabled, take 2.

This fixes the original rollback by using placeholders for the
SparseTensor shapes. The flakiness was caused by the nondeterministic
ordering of the sequences dict.

PiperOrigin-RevId: 183308774

6 years agoShow friendlier error message on failure in tf_optimizer.i
Max Galkin [Fri, 26 Jan 2018 00:08:13 +0000 (16:08 -0800)]
Show friendlier error message on failure in tf_optimizer.i

Without it we trigger a segmentation fault, but later in a different stack, which is not so helpful.

PiperOrigin-RevId: 183307729

6 years agoIn the TF cost model, ensure when id grows, records for each op's outputs match the...
Yuefeng Zhou [Thu, 25 Jan 2018 23:59:12 +0000 (15:59 -0800)]
In the TF cost model, ensure when id grows, records for each op's outputs match the number of outputs.

PiperOrigin-RevId: 183306267

6 years agoImplement LoggingAsync for GRPC Worker Services (#14604)
Sayed Hadi Hashemi [Thu, 25 Jan 2018 23:32:44 +0000 (17:32 -0600)]
Implement LoggingAsync for GRPC Worker Services (#14604)

* Implement LoggingAsync for GRPC Worker.

* Add nullptr checks

* Fix BUILD file format

* Change LoggingAsync implementation

- Revert changes to *_rendezvous_mgr
- Implement logging primitives in session_mgr instead
- Implement ClearLogs
- Fixed C++ formating

* Check for nullptr

* Better handling of the case when both "clear" and "retrieve" flags are sent to "LogAsync."
nullptr check on default_worker_cache_.

* Fix formatting

* Updata session_mgr.cc to address changes in 619792f

6 years agoadd URLEncode for the CopyObjectRequest of S3 Rename function (#16415)
Jian Lin [Thu, 25 Jan 2018 23:32:10 +0000 (07:32 +0800)]
add URLEncode for the CopyObjectRequest of S3 Rename function (#16415)

6 years agoAdd missing library in Dockerfile (#16417)
Ashish Kumar Ram [Thu, 25 Jan 2018 23:31:25 +0000 (00:31 +0100)]
Add missing library in Dockerfile (#16417)

The local Dockerfile does not have all the dependencies for running the exercise notebooks in udacity assignments.

6 years agoAdds loss_fn argument in remaining heads.
A. Unique TensorFlower [Thu, 25 Jan 2018 23:25:58 +0000 (15:25 -0800)]
Adds loss_fn argument in remaining heads.

PiperOrigin-RevId: 183301479

6 years ago[XLA:GPU] Implement conditional as a sequence of thunks in the GPU backend.
A. Unique TensorFlower [Thu, 25 Jan 2018 23:01:30 +0000 (15:01 -0800)]
[XLA:GPU] Implement conditional as a sequence of thunks in the GPU backend.

This also includes the following fixes:

(1) Update buffer assignment for conditionals so that the buffers corresponding to the true operand and the true computation parameter are colocated, and similarly, the buffers corresponding to the false operand and the false computation parameter are colocated.

(2) Update GPU copy insertion pass to insert copies when constants appear as operands of conditional instructions.

PiperOrigin-RevId: 183297282

6 years ago[XLA] Add HLO matcher for CustomCall that accepts a call target.
Justin Lebar [Thu, 25 Jan 2018 22:56:38 +0000 (14:56 -0800)]
[XLA] Add HLO matcher for CustomCall that accepts a call target.

PiperOrigin-RevId: 183296506

6 years agoWhitelist "bool" as a valid TPU infeed type.
Peter Hawkins [Thu, 25 Jan 2018 22:53:37 +0000 (14:53 -0800)]
Whitelist "bool" as a valid TPU infeed type.

PiperOrigin-RevId: 183296017

6 years agoDrop the manually_create field from RnnState.
A. Unique TensorFlower [Thu, 25 Jan 2018 22:42:03 +0000 (14:42 -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: 183294102

6 years agoUse LookupOrCreateResource when creating summary writers. The resource may
Jonathan Hseu [Thu, 25 Jan 2018 22:39:44 +0000 (14:39 -0800)]
Use LookupOrCreateResource when creating summary writers. The resource may
already exist because it's associated with the device.

PiperOrigin-RevId: 183293761

6 years agoAdd OPENSOURCE extension to schema_generated.h
A. Unique TensorFlower [Thu, 25 Jan 2018 22:38:57 +0000 (14:38 -0800)]
Add OPENSOURCE extension to schema_generated.h

PiperOrigin-RevId: 183293637

6 years agoMake select_and_scatter_test optonly
Yunxing Dai [Thu, 25 Jan 2018 22:30:15 +0000 (14:30 -0800)]
Make select_and_scatter_test optonly

- Make select_and_scatter_test optonly.
- Reduce number of shards.

PiperOrigin-RevId: 183292230

6 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

6 years agomake label_image for tflite build again (#16206)
freedom" Koan-Sin Tan [Thu, 25 Jan 2018 22:24:23 +0000 (06:24 +0800)]
make label_image for tflite build again (#16206)

* make label_image for tflite build again

1. add namespace to label_image.h to make label_image for tflite build again
2. add --config monolithic and mention NDK settings in label_image.md
3. fix a typo in display_usage()

* relies on tensor type info to switch types

use tensor types of input and output tensors instead of command
line flag from user

* reformatted according to review

6 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

6 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

6 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

6 years agoRemove calculation of unnecessary matrix columns in SVD gradient (#15801)
Viraj Navkal [Thu, 25 Jan 2018 21:11:40 +0000 (13:11 -0800)]
Remove calculation of unnecessary matrix columns in SVD gradient (#15801)

* remove calculation of unnecessary matrix columns in svd gradient

* simplify expression for svd gradient when compute_uv=False

* assign Operation attribute to local variable

6 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

6 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

6 years agoDon't load libcupti.so from regular path on Android (#16303)
Andrew Harp [Thu, 25 Jan 2018 19:56:22 +0000 (14:56 -0500)]
Don't load libcupti.so from regular path on Android (#16303)

* Don't load libcupti.so from regular path on Android

* replace NVIDIA_TEGRA with ANDROID_TEGRA to be less redundant and more specific

6 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

6 years agoAdd checkpoint file prefix check (#14341)
ted chang [Thu, 25 Jan 2018 19:16:27 +0000 (11:16 -0800)]
Add checkpoint file prefix check (#14341)

Additionally fix Two failed tests caused by the PR

Fixes #9465

6 years agoAdd additional argument to freeze_graph (#15906)
ted chang [Thu, 25 Jan 2018 19:05:08 +0000 (11:05 -0800)]
Add additional argument to freeze_graph (#15906)

6 years agofix typos (#16384)
ManHyuk [Thu, 25 Jan 2018 18:52:06 +0000 (03:52 +0900)]
fix typos (#16384)

6 years agoModified Implementation of ndlstm_base_dynamic. (#16402)
Jerome [Thu, 25 Jan 2018 18:41:30 +0000 (02:41 +0800)]
Modified Implementation of ndlstm_base_dynamic. (#16402)

* Added ctc_loss_dense_labels. This does the conversion of dense labels into sparse ones to be passed into the core ctc_loss function.

* Removed constant_op from the import.

* Matched ctc_loss_dense_labels with the other layers ops.

* Added ctc_loss_dense_labels to contrib.layers __init__.py file

* Added missing comma to list of ops.

* Reordred arguments for ctc_loss_dense_labels

Labels should be first then inputs for ctc_loss.

* Removed ctc_loss_dense_labels.

Replaced it with dense_to_sparse instead so that there'll be only one ctc_loss function.

* Replaced ctc_loss_dense_labels with dense_to_sparse

* Fixed dense_to_sparse. Some of the names of the variables did not match with that of the parameters.

* Updated documentation for dense_to_sparse since it can accept a tensor of any shape.

* Added test case for dense_to_sparse.

* Updated documentation. Dense to sparse accepts int tensors.

* Fixed testDenseFromConstantToSparse.

The sparse_to_dense order of arguments in the test are wrong and the expected constant should be of int64.

* Modified implementation of ndlstm_base_dynamic.

It now uses a BasicLSTMCell that has state_is_tuple=True to address deprecation. Right now it is still unknown why it was set to false in the first place.

6 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

6 years agoBuild libjpeg-turbo ALTIVEC SIMD (#16409)
Sam Matzek [Thu, 25 Jan 2018 18:39:31 +0000 (12:39 -0600)]
Build libjpeg-turbo ALTIVEC SIMD (#16409)

The libjpeg-turbo package has ALTIVEC SIMD and this updates the
third_party build to build the ALTIVEC SIMD on the appropriate
platform.

6 years agoimprove comment describing segmentation algorithm
David Goodwin [Thu, 25 Jan 2018 18:36:31 +0000 (10:36 -0800)]
improve comment describing segmentation algorithm