aboutsummaryrefslogtreecommitdiffhomepage
Commit message (Collapse)AuthorAge
* Add TensorArrayGradWithShape op.Gravatar A. Unique TensorFlower2018-06-08
| | | | PiperOrigin-RevId: 199862180
* Replace cout with VLOG(2).Gravatar A. Unique TensorFlower2018-06-08
| | | | PiperOrigin-RevId: 199859711
* [tf.data] Print an actionable warning when a lookup table is created in a ↵Gravatar Derek Murray2018-06-08
| | | | | | function. PiperOrigin-RevId: 199859228
* [INTEL MKL] Enable compilation of TF without MKL ML dependencyGravatar AG Ramesh2018-06-08
| | | | | | Closes #19808. PiperOrigin-RevId: 199857219
* Allow large allocations in toco.Gravatar A. Unique TensorFlower2018-06-08
| | | | PiperOrigin-RevId: 199855838
* Bugfix for dilated_conv optimizations. We were failing to create im2col ↵Gravatar A. Unique TensorFlower2018-06-08
| | | | | | arrays for dilated unstrided 1x1 cases. PiperOrigin-RevId: 199849200
* Avoid compilation of nodes that forward tensor refs.Gravatar A. Unique TensorFlower2018-06-08
| | | | PiperOrigin-RevId: 199846447
* Optimizing transpose_conv.Gravatar A. Unique TensorFlower2018-06-08
| | | | PiperOrigin-RevId: 199839745
* Change gRPC include directory from "grpc++" to "grpcpp"Gravatar A. Unique TensorFlower2018-06-08
| | | | PiperOrigin-RevId: 199836336
* Add kGenerateToken HLO instruction.Gravatar Mark Heffernan2018-06-08
| | | | | | | | | | The new HLO instruction serves two purposes. (1) It generates a new token value. This is the only way to create tokens. (2) The operation is variadic, taking zero or more token operands. The operation acts as a join of its operands. I considered initially using a kConstant constant as a method to create new tokens, but this ran into problems because of expectations in backends regarding constants and their materialization. This CL enables creation of generate-token instructions, but the new instruction is not supported yet in any backend. PiperOrigin-RevId: 199836205
* Ask NumPy for read only array when converting it to Tensor.Gravatar Eugene Zhulenev2018-06-08
| | | | | | | | | | | | | | Fix for: #17315 If numpy array is read-only, calling PyArray_FromAny with NPY_ARRAY_CARRAY flags introduce extra memory copy. Before: feed_cpu_variable_read_only: 5.6 GB/sec, min: 17.99, median: 19.54, mean: 19.76 After: feed_cpu_variable_read_only: 13.2 GB/sec, min: 7.60, median: 7.78, mean: 8.13 PiperOrigin-RevId: 199835695
* [tf.data] Adding optimization for rewriting `shuffle(...).repeat(...)` to ↵Gravatar Jiri Simsa2018-06-08
| | | | | | `shuffle_and_repeat(...)`. PiperOrigin-RevId: 199834400
* Fix: Add back test case to test generator methods.Gravatar Pavithra Vijay2018-06-08
| | | | PiperOrigin-RevId: 199834091
* Fix: Keras models using datasets in eager mode fail on float64 dataGravatar Pavithra Vijay2018-06-08
| | | | PiperOrigin-RevId: 199833632
* Improve tfdbg documentation regarding high-level APIsGravatar Shanqing Cai2018-06-08
| | | | | | | | * Mention both keras and tf.keras * In one of the early paragraphs, list all three high-level APIs supported (tf.estimator, keras and tf.contrib.slim). PiperOrigin-RevId: 199830255
* [TF:XLA] Bump open source llvm revision to r334273Gravatar Sanjoy Das2018-06-08
| | | | PiperOrigin-RevId: 199826723
* Use the new operators for list conversion. Includes list creation, append, ↵Gravatar Dan Moldovan2018-06-08
| | | | | | pop, stack. Simplify the type annotation mechanism by having it literally copy its arguments, instead of attempting to resolve them. PiperOrigin-RevId: 199822771
* Delete deprecated protos.Gravatar A. Unique TensorFlower2018-06-08
| | | | PiperOrigin-RevId: 199822232
* Split out opcodes using dimensions_ as subclasses from HloInstruction.Gravatar A. Unique TensorFlower2018-06-08
| | | | PiperOrigin-RevId: 199821675
* Preserve input shape information when serializing deferred-build Sequential ↵Gravatar Francois Chollet2018-06-08
| | | | | | models. PiperOrigin-RevId: 199817660
* Increase relative error to 1e-4 on convolution_test.Gravatar Mark Heffernan2018-06-08
| | | | | | | convolution_test had a zero relative error bound which made it overly sensitive to changes to the underlying computation. PiperOrigin-RevId: 199814523
* Automated g4 rollback of changelist 199308328Gravatar Dimitris Vardoulakis2018-06-08
| | | | PiperOrigin-RevId: 199809082
* [XLA:GPU] Add a mulit-output fusion pass to fuse sibling reduce instructions.Gravatar Thomas Joerg2018-06-08
| | | | | | Stop creating pre-fused nodes in BatchNormExpander. PiperOrigin-RevId: 199807585
* [tf.data] tf.contrib.data.CsvDataset: Add recovery for errors with quoted fieldsGravatar Rachel Lim2018-06-08
| | | | PiperOrigin-RevId: 199807061
* [tf.data] Improve the error message for `Dataset.padded_batch()`.Gravatar Derek Murray2018-06-08
| | | | | | | | | Previously, we accepted the `padded_shapes` argument without validating that it was compatible with the `input_dataset.output_shapes`. In many cases, we have enough static shape information to do this, and so we now raise an actionable error at the point where the mistake is committed, rather than at runtime. PiperOrigin-RevId: 199800348
* Fix a typo in toco flags description.Gravatar A. Unique TensorFlower2018-06-08
| | | | PiperOrigin-RevId: 199795176
* Obtain use_locking for resource variables in scatter_nd_add.Gravatar Adria Puigdomenech2018-06-08
| | | | PiperOrigin-RevId: 199782188
* Detect configurations that would be hitting bugs in cuDNN and report an error.Gravatar A. Unique TensorFlower2018-06-08
| | | | PiperOrigin-RevId: 199780350
* Fix RemoveUnusedNodes generating invalid graphs for PlaceholderWithDefault ↵Gravatar A. Unique TensorFlower2018-06-08
| | | | | | inputs PiperOrigin-RevId: 199776409
* Removes error message from queues in eager (leaves the one in queuerunners).Gravatar Alexandre Passos2018-06-08
| | | | | | | There's no real reason to not support queues in eager for people using them without using queue runners. PiperOrigin-RevId: 199770626
* Make tfe.py_func once differentiable.Gravatar Akshay Agrawal2018-06-08
| | | | | | With this change, it is now possible to embed differentiable eager code --- running on either CPU or GPU --- in graphs. Higher-order derivatives are not yet supported. PiperOrigin-RevId: 199768301
* [XLA] Base class for fusing sibling instructions with multiple outputs.Gravatar Thomas Joerg2018-06-08
| | | | PiperOrigin-RevId: 199765487
* Add checkpointing support for ReshufflingDataset.Gravatar Saurabh Saxena2018-06-07
| | | | | | This allows checkpointing input pipelines with .shuffle(reshuffle_each_iteration=True[default]) and .list_files(). PiperOrigin-RevId: 199753836
* Upgrade Eigen version. Remove eigen_fix_cuda_compilation.patch because the ↵Gravatar A. Unique TensorFlower2018-06-07
| | | | | | | | fixes in the patch have been incorporated into the Eigen opensource repository with this commit: https://bitbucket.org/eigen/eigen/commits/60ab50654998f1cbe2791d49fea94d0ca5ae08a8 PiperOrigin-RevId: 199749536
* Wire in the kDomain infrastructure brought in by cl/193798254.Gravatar A. Unique TensorFlower2018-06-07
| | | | PiperOrigin-RevId: 199745064
* Enhance row reduction implementation.Gravatar Bixia Zheng2018-06-07
| | | | | | | | | | | | | | | | | | | | The current implementation tiles the x-dimension of the tensors to calculate the partial results of the reduction. This change increases such an x-tile size from 8 to 64 if doing so results in all saturated tiles. Otherwise, this change adds z-dimension tiles to increase the number of elements that each thread reduces to a partial result to reduce the number of needed dynamic atomic operations and intra-warp reduction operations. Use a tighter yet safe loop bound for the last unsaturated tile. Avoid generating the atomic operation when the tile size is not smaller than the reduction width. Extend ForLoop emitter to support a request for fully loop unrolling. Add three tests. PiperOrigin-RevId: 199744209
* Replace add_variable() with add_weight() in official keras layers.Gravatar James Qin2018-06-07
| | | | | | Make it easier for analysis and code search. PiperOrigin-RevId: 199736646
* Implement Log operator.Gravatar A. Unique TensorFlower2018-06-07
| | | | PiperOrigin-RevId: 199735191
* [DataFlowAnalysis] Be less conservative on loop fusion nodes when reusing ↵Gravatar Yunxing Dai2018-06-07
| | | | | | | | | | buffer. - Previously, we say we cannot reuse operand buffer for a loop fusion node if any of the fusion's inputs is a broadcast or reshape. That's too conservative since in theory we can still reuse the operand's buffer if all the users of that particular operand are elementwise. This CL implements that. - Also fixed a bug in previous code where a dynamic update fusion node that ends with convert (added for bf16) is not caught by the if condition currectly. PiperOrigin-RevId: 199731488
* Changes the batch_function decorator implementation to use the newly added ↵Gravatar Vinu Rajashekhar2018-06-07
| | | | | | | | BatchFunction op. o Renames the previous version to batch_function_v1. PiperOrigin-RevId: 199729701
* Java: Release 1.9.0-rc0 (and update protbuf dependency)Gravatar Asim Shankar2018-06-07
| | | | PiperOrigin-RevId: 199729533
* Pass checkpoint_path to predicate functions for experiment.continuous_eval ↵Gravatar A. Unique TensorFlower2018-06-07
| | | | | | even in the case of falsy eval_results PiperOrigin-RevId: 199728382
* Go: Update generated wrapper functions for TensorFlow ops.Gravatar A. Unique TensorFlower2018-06-07
| | | | PiperOrigin-RevId: 199726426
* [XLA] Handle kSlice correctly in HloCostAnalysisGravatar David Majnemer2018-06-07
| | | | | | Slice doesn't read the entire input. It only reads enough to make the output. PiperOrigin-RevId: 199722987
* Update ops-related pbtxt files.Gravatar A. Unique TensorFlower2018-06-07
| | | | PiperOrigin-RevId: 199722844
* C API: Fail gracefully if the serialized graph would be too large.Gravatar Asim Shankar2018-06-07
| | | | | | | | | | | See #19657 for some motivation. Without this explicit check, a large graph would trigger an assertion failure in the protobuf codebase (https://github.com/google/protobuf/blob/0456e269ee6505766474aa8d7b8bba7ac047f457/src/google/protobuf/message_lite.cc#L68) Pull Request for google/protobuf: https://github.com/google/protobuf/pull/4739 PiperOrigin-RevId: 199719082
* [data-stats] Adds support to collect `features` and `feature-values` ↵Gravatar Shivani Agrawal2018-06-07
| | | | | | | | statistics from `Example` record of dataset. This change-list also applies transformation function `feature_stats()` to collect stats in an associated stats_aggregator (if any) to dataset in `make_batched_feature_dataset()` by default. PiperOrigin-RevId: 199718439
* TOCO: return Status instead of crashing while converting "Conv".Gravatar A. Unique TensorFlower2018-06-07
| | | | PiperOrigin-RevId: 199714511
* Split out HloFftInstruction and HloSendRecvInstruction as subclasses from ↵Gravatar A. Unique TensorFlower2018-06-07
| | | | | | HloInstruction. PiperOrigin-RevId: 199712253
* Add ScaleTriL Bijector to enable transformed distributions over PSD matrices.Gravatar A. Unique TensorFlower2018-06-07
| | | | PiperOrigin-RevId: 199706732
generated by cgit v1.2.3 (git 2.39.1) at 2024-07-05 03:24:48 +0000