aboutsummaryrefslogtreecommitdiffhomepage
Commit message (Collapse)AuthorAge
...
* | | | | | | | | | | Ensure models can be re-run with Keras (useful for Colab).Gravatar Russell Power2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Adjust TPUDistributionStrategy API to remove duplicate master address. PiperOrigin-RevId: 209058959
* | | | | | | | | | | Add a ReduceWindow test with a large window.Gravatar Michael Kuperstein2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 209056517
* | | | | | | | | | | Convert TensorFlow's flatbuffers dependency to new third party import method.Gravatar Austin Anderson2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 209056244
* | | | | | | | | | | Added a factory for StopAtCheckpointStepHook. Chief is responsible to save ↵Gravatar Mustafa Ispir2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | the checkpoint. Therefore StopAtCheckpointStepHook should not be used in chief mode. PiperOrigin-RevId: 209051351
* | | | | | | | | | | Allow tfe.set_server_def to enable remote eager execution.Gravatar Akshay Modi2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Allows a change any time within process. PiperOrigin-RevId: 209051119
* | | | | | | | | | | Improve gather ergonomics by renaming fields.Gravatar Sanjoy Das2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This CL renames the various inputs to the Gather HLO to be more mnemonic by making it more obviously a batch dynamic-slice. The replacements I made are: s/elided_window_dims/collapsed_slice_dims/g s/window_bounds/slice_sizes/g s/gather_dims_to_operand_dims/start_index_map/g s/gather_indices/start_indices/g s/output_window_dims/offset_dims/g PiperOrigin-RevId: 209051067
* | | | | | | | | | | Added non-std string support to tensorflow::StringPiece string conversion ↵Gravatar A. Unique TensorFlower2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | operator. PiperOrigin-RevId: 209050419
* | | | | | | | | | | Fix zeros_like for TensorLists.Gravatar Saurabh Saxena2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 209046988
* | | | | | | | | | | Wrap the eagerly executed bits of unit tests in their own graphsGravatar Allen Lavoie2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Provides a bit more isolation. Now with fewer reference cycles. Factors out the ResourceVariable temporary graph destruction logic into a utility so the temporary graph-building-graph doesn't bother reference cycle tests. PiperOrigin-RevId: 209041000
* | | | | | | | | | | Add a flag to crash on cuDNN cross checking failures.Gravatar Tim Shen2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 209039444
* | | | | | | | | | | Add multi_worker_util which has normalize_cluster_spec and is_chief.Gravatar Yuefeng Zhou2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 209037977
* | | | | | | | | | | Merge pull request #21324 from Intel-tensorflow:conv3dGravatar TensorFlower Gardener2018-08-16
|\ \ \ \ \ \ \ \ \ \ \ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 209032082
* | | | | | | | | | | | Add out_status to py_record_writer.writeGravatar A. Unique TensorFlower2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Throw in TFRecordWriter.write() if write fails PiperOrigin-RevId: 209030158
* | | | | | | | | | | | Add HostConstOp dependency to broken tests.Gravatar Gunhan Gulsoy2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 209029103
* | | | | | | | | | | | Refactoring: Extract IsEagerOp function to shared component.Gravatar Yu-Cheng Ling2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 209027338
* | | | | | | | | | | | Use distribution strategy to configure distribute coordinator.Gravatar Yuefeng Zhou2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Add session_creator and a couple properties to worker context which then are used to configure monitored sessions. PiperOrigin-RevId: 209026599
* | | | | | | | | | | | Fix system metadata retrieval for non-resolvable addresses within TPU ↵Gravatar Frank Chen2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Distribution Strategy PiperOrigin-RevId: 209025435
* | | | | | | | | | | | Remove magic-doc-links from code.Gravatar Mark Daoust2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This change contains no code changes. Only doc-strings. We can't use relative links in code files, so we don't have much choice but to link to tensorflow.org/ The deleted links were to docs that no longer exist. PiperOrigin-RevId: 209019572
* | | | | | | | | | | | [C API/Eager]: Fix bug in TFE_OpSetAttrString.Gravatar Asim Shankar2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | TFE_OpSetAttrString was holding on to the 'value' pointer after it returned. This bug was introduced in commit 2b0805301e4531dd7c2ed677d932f6408675460e which caused TFE_OpSetAttrString to invoke AttrBuilder& AttrBuilder::Set(StringPiece attr_name, StringPiece&& value); instead of: AttrBuilder& AttrBuilder::Set(StringPiece attr_name, T&& value) (where the latter copies 'value' when T is a StringPiece or const char* and the former aliases the memory pointed to by StringPiece). In this process, I realized that AttrBuilder::Set(StringPiece attr_name, StringPiece&& value) was never being invoked (other than in this buggy situation), so I removed it altogether. Without the changes to attr_builder.{h,cc}, the newly added test fails - complaining that "NHWC" is not a valid value for the "padding" attribute. PiperOrigin-RevId: 209017110
* | | | | | | | | | | | Do not access .op and attrs of EagerTensor in mutex.Gravatar Alexandre Passos2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 209016800
* | | | | | | | | | | | Automated rollback of commit ec5f4771e42972c31faaa39354d785891de9f91dGravatar Andrew Selle2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 209016586
* | | | | | | | | | | | Renamed function in freeze_graph.py.Gravatar Nupur Garg2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 209013414
| | | | * | | | | | | | fix feedback/copybara failure: change to self.get_temp_dir for saving checkpointGravatar weidankong2018-08-16
| | | | | | | | | | | |
* | | | | | | | | | | | NEON optimizations for second-most-common Mul uint8 kernel.Gravatar A. Unique TensorFlower2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 209010898
* | | | | | | | | | | | [Keras / Cloud TPU] Fix ClusterSpec propagation bug.Gravatar Brennan Saeta2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 209010536
* | | | | | | | | | | | Fixes issue where gradients are being quantized. In addition, because ↵Gravatar A. Unique TensorFlower2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | gradients are quantized, quantization fails for any node that is a producer of gradients since the _FollowedByFakeQuant method prevents any node followed by a fake quant from being properly quantized. PiperOrigin-RevId: 209010415
* | | | | | | | | | | | Remove magic links from docs.Gravatar Mark Daoust2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | I patched the doc generator to generate markdown links. Ran the doc converter, and copied the output into docs_src. PiperOrigin-RevId: 209010351
* | | | | | | | | | | | Enable floating-point fast-path reductions in the evaluator for types other ↵Gravatar Mark Heffernan2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | than float. Previously, reductions of non-float floating point types would check fail at runtime in the arg_literal.Get<float>() call. PiperOrigin-RevId: 209009883
* | | | | | | | | | | | internal change.Gravatar A. Unique TensorFlower2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 208999824
* | | | | | | | | | | | Add documentation clarifying the differences between tf.fill and tf.constant.Gravatar A. Unique TensorFlower2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 208998330
* | | | | | | | | | | | Move stats collection presentation-layer logic inside NodeExecStatsWrapper.Gravatar Derek Murray2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This change removes the dependency between Executor and the concrete representation of NodeExecStats as a protobuf object, is a step towards supporting multiple representations. PiperOrigin-RevId: 208995803
* | | | | | | | | | | | Support GREATER_OR_EQUAL and GREATER_THAN splits in tensor_forest evaluationGravatar A. Unique TensorFlower2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 208992685
* | | | | | | | | | | | Fixed HloSharding::Hash for tuple.Gravatar A. Unique TensorFlower2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 208991904
* | | | | | | | | | | | Don't let 0-sized arrays trip the allocator.Gravatar A. Unique TensorFlower2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 208978749
* | | | | | | | | | | | Automated rollback of commit 26353f9b51091312e7097143aee9c2d05e2011fdGravatar A. Unique TensorFlower2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 208973995
* | | | | | | | | | | | Delete tf.contrib.kfac. K-FAC in Tensorflow is now its own separate package.Gravatar Vikram Tankasali2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 208972923
* | | | | | | | | | | | Add a HloPass to rewrite convolutions with feature_group_count > 1.Gravatar Adrian Kuegel2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The convolutions are rewritten to convolutions with feature_group_count = 1. Adapt hlo_verifier and shape_inference to handle feature_group_count > 1. Add the new pass to the backends, and write a depthwise convolution test. PiperOrigin-RevId: 208954204
* | | | | | | | | | | | compat: Update forward compatibility horizon to 2018-08-16Gravatar A. Unique TensorFlower2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 208952543
* | | | | | | | | | | | Go: Update generated wrapper functions for TensorFlow ops.Gravatar A. Unique TensorFlower2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 208951388
* | | | | | | | | | | | Add a feature_group_size parameter to the Convolution HLO op.Gravatar Adrian Kuegel2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This is a first step towards supporting grouped convolutions, which are a generalization of depthwise convolution. PiperOrigin-RevId: 208950311
* | | | | | | | | | | | Update ops-related pbtxt files.Gravatar A. Unique TensorFlower2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 208949131
* | | | | | | | | | | | Add StringLength op to TensorFlow.Gravatar Sung Jin Hwang2018-08-16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This is a simple op that takes an array of strings and returns an array of the same shape, containing each string length. PiperOrigin-RevId: 208944992
* | | | | | | | | | | | Step_fn should be able to receive un-wrapped inputsGravatar Sourabh Bajaj2018-08-15
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 208929959
* | | | | | | | | | | | Separated protobuf_compiler into a separate BUILD rule.Gravatar A. Unique TensorFlower2018-08-15
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This prevents most of TensorFlow from needing to depend on libprotoc. PiperOrigin-RevId: 208929776
* | | | | | | | | | | | Require token operand for infeed and outfeed.Gravatar Mark Heffernan2018-08-15
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | For expediency in rolling out support for tokens used for ordering side-effecting ops, infeed and outfeed *optionally* took a token operand. This CL removes that option so all infeed and outfeed instructions take a token operand. PiperOrigin-RevId: 208927968
* | | | | | | | | | | | Update the ADC file location in Colab.Gravatar Craig Citro2018-08-15
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 208925100
* | | | | | | | | | | | In auto-tuning, initialize fp16 buffers with 0.1.Gravatar Tim Shen2018-08-15
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 208918032
| | | | | | | | * | | | Fix a build issueGravatar Mahmoud Abuzaina2018-08-15
| |_|_|_|_|_|_|/ / / / |/| | | | | | | | | |
* | | | | | | | | | | Allow python3.5 devel docker builds.Gravatar Amit Patankar2018-08-15
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 208914243
* | | | | | | | | | | LARS Optimizer in TensorFlowGravatar A. Unique TensorFlower2018-08-15
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Based on CL from Chris Ying and contributions from Y. You and Wang Tao. Introduced by "Large Batch Training of Convolutional Networks" by Y. You, I. Gitman, and B. Ginsburg. (https://arxiv.org/abs/1708.03888) Implements the LARS learning rate scheme presented in the paper above. This optimizer is useful when scaling the batch size to up to 32K without significant performance degradation. It is recommended to use the optimizer in conjunction with: - Gradual learning rate warm-up - Linear learning rate scaling - Poly rule learning rate decay With this optimizer, ResNet-50 now converges to 76.3% top-1 accuracy at batch 32K on a JF Pod. PiperOrigin-RevId: 208914187
generated by cgit v1.2.3 (git 2.39.1) at 2024-07-06 01:41:01 +0000