aboutsummaryrefslogtreecommitdiffhomepage
Commit message (Collapse)AuthorAge
...
* [TF:XLA] Create Despecializing Pass PipelineGravatar Nick Desaulniers2018-04-06
| | | | | | | | When comparing backends, it is useful to take an HLO optimized for one backend and perform transformations in order to match numerics. This can be thought of as finding a lowest common denominator. Move this grouping of passes into its own HloPassPipeline that can be reused in a few different places. PiperOrigin-RevId: 191914799
* Added headers only version of tensorflow/core/kernels:cwise_lib, ↵Gravatar A. Unique TensorFlower2018-04-06
| | | | | | cwise_lib_hdrs. This is for clients that want to use the cwise_ops machinery when making their own custom ops, including cwise_lib directly causes multiple definition linker errors. PiperOrigin-RevId: 191914445
* Minor doc clarification for reduce_sum return typeGravatar A. Unique TensorFlower2018-04-06
| | | | PiperOrigin-RevId: 191914398
* Properly handle callable objects.Gravatar A. Unique TensorFlower2018-04-06
| | | | PiperOrigin-RevId: 191913834
* TPU Cost Estimator has been modified to also account for the memory cost in ↵Gravatar A. Unique TensorFlower2018-04-06
| | | | | | the execution time. Until more sophisticated methods are added, we resort to the roofline model to calculate such cost. PiperOrigin-RevId: 191913626
* Change GetInstructionCallContext to take an opcode instead of anGravatar Mark Heffernan2018-04-06
| | | | | | | | HloInstruction. This enables use of the function without an actual instruction (eg, if you just have an HloProto). PiperOrigin-RevId: 191905914
* Allow TFE_NewContext to fail more reasonably when SWIG is checking status.Gravatar Akshay Modi2018-04-06
| | | | | | | | | | | | | | | Before: TFE_Context would check nullptr, and the function would fail straight away. Now: TFE_Context is nullptr, so it skips down to checking the status, and an error is raised. I'm not able to find in SWIG documentation how to order typemaps in the generated code - ideally, I'd order it to check the status typemap first. This code makes it not dependent on this ordering either way. PiperOrigin-RevId: 191905893
* Object-based checkpointing support for unidirectional cuDNN LSTM cellsGravatar Allen Lavoie2018-04-06
| | | | | | | | | | | | | | | | Once checked in, this will be the only way I know of to save canonical weights when executing eagerly. Eager's name-based saving support will only do the opaque parameter buffer. I'm not going to try converting everything in one go, but it's a start at least. And everything else should raise a NotImplementedError rather than silently not saving correctly. Single-layer cuDNN cells can be swapped for un-wrapped cuDNN compatible cells or single cells wrapped in MultiRNNCells. Multi-layer cells need MultiRNNCell wrapping. PiperOrigin-RevId: 191905703
* Materialize tensor array sizes whenever possibleGravatar Benoit Steiner2018-04-06
| | | | PiperOrigin-RevId: 191900015
* [tf.data] Enable using `tf.contrib.data.prefetch_to_device()` in eager mode.Gravatar Derek Murray2018-04-06
| | | | | | | | | The added functionality is a substitute for the implicit prefetching in `tfe.Iterator`, and the two paths will converge in a future change. Fixes #18260. PiperOrigin-RevId: 191897666
* [TPUClusterResolver] Start a TFServer when running in GKEGravatar Brennan Saeta2018-04-06
| | | | | | This change allows advanced input pipelines (e.g. StreamingFilesDataset, or split-pipelines that use py_func's) to run in GKE- and GKE-like enviornments. PiperOrigin-RevId: 191897639
* Tweaked docstrings in LayerCollection.Gravatar A. Unique TensorFlower2018-04-06
| | | | PiperOrigin-RevId: 191897098
* Support override of device filters for gRPC, by overriding the requests with ↵Gravatar A. Unique TensorFlower2018-04-06
| | | | | | default session config. PiperOrigin-RevId: 191895856
* Fix up the support for the case where a given array name occurs multipleGravatar A. Unique TensorFlower2018-04-06
| | | | | | | | | times in the inputs/outputs list of an op. The (non-essential) computation of the optimal workspace size had not been updated for that case, causing it to fail on a simple test case. Moreover, the initial implementation had some redundant usage of std::find that this CL moves to a shared helper function. PiperOrigin-RevId: 191894081
* The training model need not be built when the kfac optimizer is initialized ↵Gravatar A. Unique TensorFlower2018-04-06
| | | | | | | | so the self._variables will be empty list. So pass a function which returns list of trainable variables to estimator. PiperOrigin-RevId: 191893084
* Add description to the LPIRC 2018 competition benchmarker.Gravatar A. Unique TensorFlower2018-04-06
| | | | PiperOrigin-RevId: 191889484
* Avoid marking clusters containing only Identity ops for compilation.Gravatar A. Unique TensorFlower2018-04-06
| | | | | | This would produce clusters where XLA cannot optimize anything. PiperOrigin-RevId: 191887414
* Define PRNG seeding style for new code in Distributions and TF Probability, ↵Gravatar A. Unique TensorFlower2018-04-06
| | | | | | | | | | with rationales. Implement lightweight PRNG for seed generation in that style. Enables incremental refactoring of existing code into this style. PiperOrigin-RevId: 191884573
* Fix typos in XlaCompilationCacheGravatar Igor Ganichev2018-04-06
| | | | PiperOrigin-RevId: 191881135
* internal changeGravatar A. Unique TensorFlower2018-04-06
| | | | PiperOrigin-RevId: 191869400
* Add a test to check graceful handling of out-of-memory conditions.Gravatar A. Unique TensorFlower2018-04-06
| | | | PiperOrigin-RevId: 191860462
* Fix StringPiece use-after-free in MasterSession::ReffedClientGraph.Gravatar Derek Murray2018-04-05
| | | | | | Use the owned ClientGraph as the source for the node_to_name_ map, rather than the borrowed GraphExecutionState (which can be deleted while the ReffedClientGraph is in use). PiperOrigin-RevId: 191847023
* Upgrade libpngGravatar Justine Tunney2018-04-05
| | | | PiperOrigin-RevId: 191840652
* Swap in the new implementation of while and for loops.Gravatar A. Unique TensorFlower2018-04-05
| | | | PiperOrigin-RevId: 191838806
* Expose the adaptive sampling option for SDCA and shuffle the data when ↵Gravatar A. Unique TensorFlower2018-04-05
| | | | | | adaptive sampling is off. PiperOrigin-RevId: 191836004
* [XLA] Don't call MultidimensionalIndexToLinearIndex in HloEvaluator's ↵Gravatar Justin Lebar2018-04-05
| | | | | | | | | | convolution routine. Before: ConvolutionTest/0.StridedFilter (41812 ms) After: ConvolutionTest/0.StridedFilter (28054 ms) Speedup: 42 / 28 = 1.5x PiperOrigin-RevId: 191835735
* Add RunMetadata logging to tf.train.ProfilerHook for Tensorboard Memeory/CPU ↵Gravatar A. Unique TensorFlower2018-04-05
| | | | | | usage visualization PiperOrigin-RevId: 191832832
* Added `drop_final_batch` argument to make_batched_features_dataset. This ↵Gravatar A. Unique TensorFlower2018-04-05
| | | | | | allows the batch_and_drop_remainder function to be used instead of the default batch function. PiperOrigin-RevId: 191831842
* [XLA] Don't call Literal::Get in HloEvaluator's convolution loop.Gravatar Justin Lebar2018-04-05
| | | | | | | | | | | | | This speeds up the implementation of conv because Literal::Get calls Literal::Piece::data, which is relatively slow. Instead, we call Literal::Data() once and cache the result. Before: ConvolutionTest/0.StridedFilter (59094 ms) After: ConvolutionTest/0.StridedFilter (41812 ms) Speedup: 59/42 = 1.4x PiperOrigin-RevId: 191830741
* Lazily evaluate shapes with the C API enabled.Gravatar Skye Wanderman-Milne2018-04-05
| | | | | | | | | | | | This change makes it so shapes are computed only when requested with _USE_C_API = True. Note that the C API will still raise a shape error if necessary when the op is created. In addition, it cleans up the logic for _USE_C_SHAPES = True. In this case, we lazily fetch and cache shapes directly from the C API. We no longer need set_shapes_for_outputs at all in this case. PiperOrigin-RevId: 191830565
* refactor and add proto field required by POD support.Gravatar A. Unique TensorFlower2018-04-05
| | | | PiperOrigin-RevId: 191826636
* Add a command line parameter to toco to change the way toco rescales input ↵Gravatar A. Unique TensorFlower2018-04-05
| | | | | | and output tensors. PiperOrigin-RevId: 191825756
* Automated g4 rollback of changelist 191605505Gravatar Tony Wang2018-04-05
| | | | PiperOrigin-RevId: 191824447
* Make concat handler support mixed range inputGravatar A. Unique TensorFlower2018-04-05
| | | | PiperOrigin-RevId: 191822664
* Added an option to inline all functions in aggressive mode.Gravatar Benoit Steiner2018-04-05
| | | | PiperOrigin-RevId: 191819577
* Add link for index file in performance tab.Gravatar Mark Daoust2018-04-05
| | | | PiperOrigin-RevId: 191811610
* Add functions to extract the basic symbols on which a composite name relies. ↵Gravatar A. Unique TensorFlower2018-04-05
| | | | | | This in turn allows to statically obtain a block's syntactic closure. PiperOrigin-RevId: 191809965
* [TF:XLA] No need to set return value in the while loop's condition.Gravatar A. Unique TensorFlower2018-04-05
| | | | PiperOrigin-RevId: 191809110
* Add for and while loops to the list of operators. Do not use them yet.Gravatar A. Unique TensorFlower2018-04-05
| | | | PiperOrigin-RevId: 191807973
* Add Raspberry Pi section and link to github build instructions.Gravatar Billy Lamberta2018-04-05
| | | | PiperOrigin-RevId: 191807862
* Make tf.contrib.estimator.add_metrics work with warm-starting.Gravatar A. Unique TensorFlower2018-04-05
| | | | PiperOrigin-RevId: 191805682
* Disable tests that are currently failing with cuda 9Gravatar A. Unique TensorFlower2018-04-05
| | | | PiperOrigin-RevId: 191805453
* [XLA] Better support for mul reductions in MakeFakeArguments()Gravatar Michael Kuperstein2018-04-05
| | | | | | Mul reductions want a 1 as their init value, not a 0 or a random value. PiperOrigin-RevId: 191802819
* Fix TF_ImportGraphDefResults and TF_Function leaks in Python API.Gravatar Skye Wanderman-Milne2018-04-05
| | | | PiperOrigin-RevId: 191797853
* quantized LSTM support improvementsGravatar A. Unique TensorFlower2018-04-05
| | | | PiperOrigin-RevId: 191794956
* Changes loss_reduction default to SUM_OVER_BATCH_SIZE for multi_class_head ↵Gravatar A. Unique TensorFlower2018-04-05
| | | | | | and binary_classification_head. PiperOrigin-RevId: 191793392
* Enable branch prediction in TensorFlowGravatar Justine Tunney2018-04-05
| | | | PiperOrigin-RevId: 191788253
* [XLA] Remove a dead function and a stale todo.Gravatar A. Unique TensorFlower2018-04-05
| | | | PiperOrigin-RevId: 191786563
* Automated g4 rollback of changelist 191753026Gravatar A. Unique TensorFlower2018-04-05
| | | | PiperOrigin-RevId: 191784709
* Save the original from_proto method before calling it to avoid infinite loop.Gravatar Priya Gupta2018-04-05
| | | | PiperOrigin-RevId: 191784430
generated by cgit v1.2.3 (git 2.39.1) at 2024-07-02 04:52:20 +0000