aboutsummaryrefslogtreecommitdiffhomepage
Commit message (Collapse)AuthorAge
* Remove _USE_C_API staging from ops.py.Gravatar Skye Wanderman-Milne2018-06-05
| | | | PiperOrigin-RevId: 199298594
* Fix generated_zip_test failure caused by regex matching failures.Gravatar A. Unique TensorFlower2018-06-05
| | | | PiperOrigin-RevId: 199296333
* Remove test dependencies that are no longer needed.Gravatar Dan Moldovan2018-06-05
| | | | PiperOrigin-RevId: 199293694
* Handle scalar input to assert_equal in eager.Gravatar Tom Hennigan2018-06-05
| | | | PiperOrigin-RevId: 199274329
* Added missing backtick in tf.ones_like documentationGravatar A. Unique TensorFlower2018-06-05
| | | | PiperOrigin-RevId: 199262414
* Windows: Refactor bazel_test_lib.sh and common_env.shGravatar A. Unique TensorFlower2018-06-05
| | | | | | | | | - Removed workaround for https://github.com/bazelbuild/bazel/issues/2182 since it's fixed - Removed setting CUDA related environment variables. Assume they are already set. If not, configure.py will set default values for them. - Removed obsolete variables for cc_test targets. PiperOrigin-RevId: 199256482
* [TF:XLA] Add method CreateNewModule to HloVerifiedTestBase, and remember all ↵Gravatar Dimitris Vardoulakis2018-06-04
| | | | | | created modules, to verify at TearDown. PiperOrigin-RevId: 199244092
* Resolve device names when passed into DistributionStrategy methods.Gravatar A. Unique TensorFlower2018-06-04
| | | | PiperOrigin-RevId: 199241723
* Enable XLA fusions as a Grappler optimization.Gravatar A. Unique TensorFlower2018-06-04
| | | | PiperOrigin-RevId: 199230907
* [XLA] Don't dump subgraphs twice in hlo_graph_dumper.Gravatar Justin Lebar2018-06-04
| | | | | | | Surprisingly a subgraph twice mostly worked. But it broke the rollover edge highlighting, and it also drew all the edges in the subgraph twice. PiperOrigin-RevId: 199221368
* HloParser: use uint16 in U16 caseGravatar Yunxing Dai2018-06-04
| | | | PiperOrigin-RevId: 199220422
* Updated include path for internal protobuf implementation.Gravatar A. Unique TensorFlower2018-06-04
| | | | PiperOrigin-RevId: 199216721
* Fix an floating point inaccuracy issue in ↵Gravatar A. Unique TensorFlower2018-06-04
| | | | | | | | | | | | | | | | | | | | | | | | precision_recall_at_equal_thresholds due to accumulating the tp/fp/tn/fn values in float32, which can become highly inaccurate as the number of values increases. In the common case, the method sums the value 1.0f to the tp/fp/tn/fn bucket for every value in the predictions tensor. If the tensor is large (say, it represents an image and we have one tp/fp/tn/fn value per pixel), then we are essentially adding many 1.0f's together, across the entire batch and also across all the batches. By doing it in float32 the value starts becoming inaccurate at around 16M, which is very small. In practice, we see a deviation of 100x when the total reaches about 3e10 (the previous code reports a number about 1e8 when the actual value should be 3e10). We avoid all these issues by always accumulating in float64. Also fix a bug that the method cannot be called with predictions dtype being anything other than float32. Preivously it would crash due to the eps code near the end. Added tests for using float64 and float16. PiperOrigin-RevId: 199216173
* Add TOKEN primitive type.Gravatar Mark Heffernan2018-06-04
| | | | | | | | The token type will be threaded through side-effecting ops to order them. Subsequent cls will add new opcodes and change side effecting operations to support this ordering. This CL also does some cleanup in shape_util and layout_util where we have assumed that shapes are either arrays or tuples. PiperOrigin-RevId: 199215963
* Add support for kDomain parsing in HLO parser.Gravatar A. Unique TensorFlower2018-06-04
| | | | PiperOrigin-RevId: 199208527
* More handle_data fixing.Gravatar Skye Wanderman-Milne2018-06-04
| | | | | | I'm not sure why our existing tests didn't catch this... PiperOrigin-RevId: 199206183
* nit: FlatBuffer -> FrozenGraphGravatar Suharsh Sivakumar2018-06-04
| | | | PiperOrigin-RevId: 199205459
* Adds update_ops to train_op for all heads.Gravatar A. Unique TensorFlower2018-06-04
| | | | PiperOrigin-RevId: 199203634
* Remove tf_export decorator from contrib. tf_export decorators currently aren'tGravatar Anna R2018-06-04
| | | | | | supported in contrib. PiperOrigin-RevId: 199200258
* Move benchmarking code to a new directory and add some documentation.Gravatar Shashi Shekhar2018-06-04
| | | | PiperOrigin-RevId: 199200246
* Add rip-offs of LLVM's cast, dyn_cast, cast_or_null, dyn_cast_or_null in ↵Gravatar A. Unique TensorFlower2018-06-04
| | | | | | preparation to split HloInstruction into subclasses. This initial implementation uses C++ dynamic_cast, so it also adds vtable to HloInstruction. PiperOrigin-RevId: 199199109
* Fixes a cleanup bug in BatchFunction op.Gravatar Vinu Rajashekhar2018-06-04
| | | | PiperOrigin-RevId: 199198413
* Add various missing aliases for symbols in tf.keras submodules.Gravatar Francois Chollet2018-06-04
| | | | PiperOrigin-RevId: 199198086
* Remove --distinct_host_configuration=false from tools/bazel.rcGravatar Amit Patankar2018-06-04
| | | | | | | | Don't use --distinct_host_configuration=false by default, because it would break cross compiling, like android build and Raspberry Pi build. Instead, we add it for builds that we know they have the same host and target platforms. PiperOrigin-RevId: 199194260
* [XLA:GPU] Propagate layouts in a better order for performance and fusion.Gravatar Blake Hechtman2018-06-04
| | | | PiperOrigin-RevId: 199193181
* Improve TOCO error handling.Gravatar Nupur Garg2018-06-04
| | | | PiperOrigin-RevId: 199186109
* Fix broken distributed_runtime/remote_device_test by adding missingGravatar A. Unique TensorFlower2018-06-04
| | | | | | std::shared_ptr. PiperOrigin-RevId: 199179607
* Fix a couple of doc typos.Gravatar A. Unique TensorFlower2018-06-04
| | | | PiperOrigin-RevId: 199179067
* Turns on optimization to convert division of sqrt to multiplication of rsqrtGravatar A. Unique TensorFlower2018-06-04
| | | | PiperOrigin-RevId: 199177029
* added clearer description for invalid behavior when executing in eager mode.Gravatar A. Unique TensorFlower2018-06-04
| | | | PiperOrigin-RevId: 199173022
* Fix Python API.Gravatar Nupur Garg2018-06-04
| | | | PiperOrigin-RevId: 199171845
* Fix test user opsGravatar Akshay Modi2018-06-04
| | | | PiperOrigin-RevId: 199171316
* [TF:XLA] Bump open source llvm revision to r333878Gravatar Sanjoy Das2018-06-04
| | | | PiperOrigin-RevId: 199168290
* Add a special functions module that contains non-Python abstractions, like ↵Gravatar Dan Moldovan2018-06-04
| | | | | | the list stack operation. PiperOrigin-RevId: 199167953
* - Fix typo in evaluatorGravatar Yunxing Dai2018-06-04
| | | | PiperOrigin-RevId: 199164433
* Fix visibility for tf.keras.__version__Gravatar Mark Daoust2018-06-04
| | | | PiperOrigin-RevId: 199161696
* Computing the volume of the set of correlation matrices with bounded ↵Gravatar A. Unique TensorFlower2018-06-04
| | | | | | | | determinant. This is useful for testing the LKJ distribution on correlation matrices. PiperOrigin-RevId: 199153115
* [TF2XLA] Change to resize bilinear to between match a BackpropInput ↵Gravatar Blake Hechtman2018-06-04
| | | | | | convolution by swapping the kernel input and output feature dimension. PiperOrigin-RevId: 199153010
* Minor error message fix in TPUEstimator.Gravatar Yanan Cao2018-06-04
| | | | PiperOrigin-RevId: 199148136
* Build TF on Windows with --config=optGravatar A. Unique TensorFlower2018-06-04
| | | | | | | | --config=opt will enable /arch:AVX cc option on Windows -c opt is already specified in tools/bazel.rc, no it's OK to remove it here PiperOrigin-RevId: 199145562
* Update the distributed SDCA test.Gravatar A. Unique TensorFlower2018-06-04
| | | | PiperOrigin-RevId: 199142338
* Enable cross-device dependency grouping optimization in non-AGGRESSIVE modes.Gravatar Peter Hawkins2018-06-04
| | | | PiperOrigin-RevId: 199141605
* Internal changeGravatar A. Unique TensorFlower2018-06-04
| | | | PiperOrigin-RevId: 199140124
* Disable flaky test ↵Gravatar Peter Hawkins2018-06-04
| | | | | | tensorflow/contrib/distribute/python:minimize_loss_test_gpu from continuous builds. PiperOrigin-RevId: 199140117
* Add LRN as unchanged rf layer operations for the receptive field calculator.Gravatar A. Unique TensorFlower2018-06-04
| | | | PiperOrigin-RevId: 199134753
* [XLA:GPU] Add error message to CHECK for preconditions to lower fusions with ↵Gravatar A. Unique TensorFlower2018-06-04
| | | | | | multiple reduce outputs. PiperOrigin-RevId: 199132442
* Temporary patch: properly handle expressions in subscripts. The long term ↵Gravatar Dan Moldovan2018-06-04
| | | | | | | | | fix is either of: (a) dropping support for tracking specific slices of a symbol (b) track slices along with the symbols on which they depend. Background: So far we tracked symbols like `a[b]` and allow conversions of the kind `if <cond>: a[b] = c` -> `a[b] = ag__.if_stmt(<cond>, lambda: c, lambda: a[b])`. That construct allowed a to be anything, including e.g. Python lists, objects. etc. This is incomplete and will in the future become obsolete as we override the slice operator. In effect the statement above will be converted to `a = ag__.if_stmt(<cond>, lambda: ag__.set_item(a, b, c), lambda: a)`. However, this latter form does not support objects, so there is a tradeoff. PiperOrigin-RevId: 199131573
* Add stored eager variables to graph collections.Gravatar Tom Hennigan2018-06-04
| | | | PiperOrigin-RevId: 199125920
* Update README.md for tf.contrib.kfac and add deprecation warning.Gravatar Vikram Tankasali2018-06-04
| | | | PiperOrigin-RevId: 199119904
* [XLA] Remove unnecessary std::vector copiesGravatar Benjamin Kramer2018-06-04
| | | | | | We can just pass along the original ArraySlice. PiperOrigin-RevId: 199109815
generated by cgit v1.2.3 (git 2.39.1) at 2024-07-04 00:54:39 +0000