aboutsummaryrefslogtreecommitdiffhomepage
Commit message (Collapse)AuthorAge
...
* Use best practices junit definitionGravatar Justine Tunney2017-02-28
| | | | Change: 148839605
* Make selective registration handle spaces in kernel name.Gravatar Manjunath Kudlur2017-02-28
| | | | Change: 148837641
* Fixing a bug in GetMatchingFiles when the top level directory is null. There ↵Gravatar Rohan Jain2017-02-28
| | | | | | was an inconsistency in which we translated the directory and all the patterns etc. that were being matched. As a result ./<pattern> worked but <pattern> did not. Change: 148836014
* Set explicit launch bounds on DepthwiseConv2dBackpropInputGPUKernel().Gravatar A. Unique TensorFlower2017-02-28
| | | | | Otherwise it ends up using too many registers and causes launch failures. Change: 148834647
* Use 'name' instead of 'test_name' when forming benchmark output file name.Gravatar A. Unique TensorFlower2017-02-28
| | | | Change: 148834478
* Refactor python BUILD targets.Gravatar Skye Wanderman-Milne2017-02-28
| | | | | | | | | | The deps of the "framework" target did not reflect the actual imports used in those files. Adding the missing deps causes circular dependencies, so this patch refactors "framework" into per-file targets. It still exposes the "framework" and "framework_for_generated_wrappers" to avoid breaking the many targets that have these as dependencies. Change: 148834459
* Fix user ops test broken by doc file moveGravatar Shanqing Cai2017-02-28
| | | | Change: 148833774
* Cleaner error messages for RNNCell scope failures.Gravatar Eugene Brevdo2017-02-28
| | | | Change: 148832474
* Fixes XLA open source build.Gravatar Blake Hechtman2017-02-28
| | | | Change: 148831489
* Add optional transport_options to RecvTensorRequest.Gravatar A. Unique TensorFlower2017-02-28
| | | | Change: 148827175
* Make TensorForest dense input shape a little more flexible by inferring size ↵Gravatar A. Unique TensorFlower2017-02-28
| | | | | | 1 if needed. Change: 148825114
* Migrate tf_scalar_dashboard to webfiles.Gravatar Dandelion Mané2017-02-28
| | | | Change: 148824590
* Update android version in Docker.Gravatar A. Unique TensorFlower2017-02-28
| | | | Change: 148823878
* Go: Update generated wrapper functions for TensorFlow ops.Gravatar A. Unique TensorFlower2017-02-28
| | | | Change: 148820986
* Update ops-related pbtxt files.Gravatar A. Unique TensorFlower2017-02-28
| | | | Change: 148820749
* Only deprecate QuantizeAndDequantize at version 22.Gravatar Suharsh Sivakumar2017-02-28
| | | | Change: 148816238
* Actually forward inputs in linear algebra ops. Previously, the Tensor copies ↵Gravatar A. Unique TensorFlower2017-02-28
| | | | | | | performed when building the "inputs" array would increment the refcount and prevent forwarding. Add a virtual method in the base class to disable forwarding for specific ops. In particular, the Eigen code used for MatrixSolveLS and MatrixTriangularSolve assumes no aliasing. Change: 148813349
* Adds const& in one place where it's possible.Gravatar Alexandre Passos2017-02-28
| | | | Change: 148809526
* Accept scalar inputs for categorical columns.Gravatar David Soergel2017-02-28
| | | | Change: 148809067
* Remove the "Legacy Underscore Categorization" option from TensorBoard.Gravatar Dandelion Mané2017-02-28
| | | | | I want to clean up the UI, and I don't think it's worth supporting this feature anymore. Change: 148807695
* examples/android/README.md: lower example image width to allow them to ↵Gravatar Andrew Harp2017-02-28
| | | | | | properly display on the same row. Change: 148806312
* Migrate vz_distribution_chart and tf_distribution_dashboard to webfiles.Gravatar Dandelion Mané2017-02-28
| | | | Change: 148806189
* Fix broken tensorflow-cl-windows-cmakeGravatar A. Unique TensorFlower2017-02-28
| | | | Change: 148805879
* When reuse is False in variable_scope, treat is as None (inherit from parent).Gravatar Lukasz Kaiser2017-02-28
| | | | | (All known callers already assumed that this was the case before.) Change: 148805391
* Remove kubectl_util and benchmark_util from tensorflow repo since they now liveGravatar A. Unique TensorFlower2017-02-28
| | | | | under benchmarks repo (https://github.com/tensorflow/benchmarks). Change: 148802196
* Use buffer forwarding in assign_op.Gravatar A. Unique TensorFlower2017-02-28
| | | | Change: 148797512
* Replace broken junit download link with maven.Gravatar A. Unique TensorFlower2017-02-28
| | | | Change: 148795638
* Removes unecessary tf_gather from cross entropy loss.Gravatar Blake Hechtman2017-02-28
| | | | Change: 148794476
* [XLA] Add predicate reduce test.Gravatar Jacques Pienaar2017-02-28
| | | | Change: 148791273
* tfdbg: add option to tolerate debug op creation failuresGravatar Shanqing Cai2017-02-28
| | | | Change: 148788753
* Ensures Bitcast is not performed between different elment sizes.Gravatar Blake Hechtman2017-02-28
| | | | | | Removes restrictive requirement that replacing an operand requires compatible shapes which is not true with broadcast optimizations. Change: 148786513
* Used scaffold in estimator.evaluate.Gravatar Mustafa Ispir2017-02-28
| | | | Change: 148786108
* Simple benchmark of resource manager overhead.Gravatar Alexandre Passos2017-02-28
| | | | Change: 148778464
* Work around bug in some Clang versions that can lead to these macros expandedGravatar A. Unique TensorFlower2017-02-28
| | | | | into different string literals compared to GCC. Change: 148767816
* Add a ScheduledOutputTrainingHelper.Gravatar A. Unique TensorFlower2017-02-28
| | | | Change: 148765793
* Go: Provide a mechanism to configure the Session.Gravatar Asim Shankar2017-02-28
| | | | | | | | A Session is configured using the ConfigProto protocol buffer. For now, continuing with attempts to keep the 'tensorflow' go package free of any protocol buffer dependencies, SessionOptions uses a serialized representation of this message. This choice might make sense to revisit. Change: 148750535
* Automated rollback of change 148702459Gravatar Gunhan Gulsoy2017-02-28
| | | | Change: 148740836
* Work around bug in some Clang versions that can lead to these macros expandedGravatar A. Unique TensorFlower2017-02-27
| | | | | into different string literals compared to GCC. Change: 148735423
* Fixing variables typo in part of the code sampleGravatar A. Unique TensorFlower2017-02-27
| | | | | And param order in nce_loss in the code sample Change: 148732811
* Users can pass a list of class ids to multiclass/multilabel heads where we ↵Gravatar Zakaria Haque2017-02-27
| | | | | | compute some per class metrics. The per class AUC had a bug where we were using logits rather than probabilities. This CL fixes this bug. Change: 148727736
* Added a test to check model_fn with lambdaGravatar Mustafa Ispir2017-02-27
| | | | Change: 148709245
* state_saving_rnn_estimator_test.py is no longer flaky.Gravatar John Bates2017-02-27
| | | | Change: 148707426
* Remove legacy special case for ToFloat in InferAllocAttr(). The ToFloat op ↵Gravatar A. Unique TensorFlower2017-02-27
| | | | | | | behaves like any other cast op now. Micro-optimization: Don't call InferAllocAttr on control edges, since the result is ignored anyway. Change: 148705319
* Add tensorflow::gtl::optional.Gravatar Justin Lebar2017-02-27
| | | | Change: 148702459
* Adding a mean per-class accuracy metric.Gravatar Dumitru Erhan2017-02-27
| | | | Change: 148700711
* [XLA] Add basic outfeed support.Gravatar Jacques Pienaar2017-02-27
| | | | Change: 148699787
* Moved common code out of RNN Estimators.Gravatar John Bates2017-02-27
| | | | Change: 148698589
* Disable evaluation_test from cmake build.Gravatar Gunhan Gulsoy2017-02-27
| | | | | It is disabled in bazel already. Change: 148691841
* Java: Memory leak fix and Windows friendliness.Gravatar Asim Shankar2017-02-27
| | | | | | | - Delete memory allocated for exception error messages. - Use vsnprintf instead of vasprintf for allocating exception error messages since the latter doesn't seem to be easily available on Windows. Change: 148687623
* Fix the estimator import error on windows cmake buildGravatar Jianwei Xie2017-02-27
| | | | Change: 148684190