aboutsummaryrefslogtreecommitdiffhomepage
Commit message (Collapse)AuthorAge
...
* Fix the python 3 breakage caused by 172073518Gravatar Gunhan Gulsoy2017-10-13
| | | | PiperOrigin-RevId: 172118528
* Make Per-Host input_fn deployment as default (instead of Per-Core)Gravatar Jianwei Xie2017-10-13
| | | | PiperOrigin-RevId: 172117003
* Fix a dtypeGravatar A. Unique TensorFlower2017-10-13
| | | | PiperOrigin-RevId: 172114960
* Removing custom CROSSTOOL and BUILD file for TF cpu builds as they are ↵Gravatar A. Unique TensorFlower2017-10-13
| | | | | | redundant now that bazel-toolchains repo is live PiperOrigin-RevId: 172113861
* Direct users of the old KMeans estimator to use the new one that's built ↵Gravatar A. Unique TensorFlower2017-10-13
| | | | | | from the core Estimator API. PiperOrigin-RevId: 172108321
* adding test for heterogeneous dataset from generatorGravatar Jiri Simsa2017-10-13
| | | | PiperOrigin-RevId: 172107872
* Internal change.Gravatar A. Unique TensorFlower2017-10-13
| | | | PiperOrigin-RevId: 172105420
* Avoid cache thrashing in CTC beam searchGravatar A. Unique TensorFlower2017-10-13
| | | | | | Change the logic that identifies topK choices with a cache friendly alternative. PiperOrigin-RevId: 172101068
* [XLA] Add ShiftLeft, ShiftRightArithmetic, and ShiftRightLogical operators.Gravatar Peter Hawkins2017-10-13
| | | | PiperOrigin-RevId: 172091595
* [TF:XLA] Add bound to ArgMax in randomized tests.Gravatar Peter Hawkins2017-10-13
| | | | PiperOrigin-RevId: 172091245
* Small improvement of verification heuristics.Gravatar A. Unique TensorFlower2017-10-13
| | | | PiperOrigin-RevId: 172073518
* Extend the ExecuteParallel service interface to allow multiple devices per ↵Gravatar A. Unique TensorFlower2017-10-13
| | | | | | computation. PiperOrigin-RevId: 172071664
* Automated g4 rollback of changelist 172048554Gravatar Anna R2017-10-13
| | | | PiperOrigin-RevId: 172065800
* Add "clear_output_shapes" option to FoldConstants transformer inGravatar A. Unique TensorFlower2017-10-12
| | | | | | | | | tools/graph_transforms. By setting this option to false, the transformer will not strip off the shape information stored as attributes. PiperOrigin-RevId: 172057283
* Make SavedModel loading forward compatible by removing default attributes ↵Gravatar Kiril Gorovoy2017-10-12
| | | | | | from the graph def that were not present in the consumer (server). PiperOrigin-RevId: 172051437
* Replace typeof with decltype.Gravatar Gunhan Gulsoy2017-10-12
| | | | PiperOrigin-RevId: 172051036
* Terminate process when we hit an unexpected exception in the infeed queue.Gravatar Russell Power2017-10-12
| | | | PiperOrigin-RevId: 172050536
* Add streaming_false_{negative,positive}_rate and ↵Gravatar A. Unique TensorFlower2017-10-12
| | | | | | streaming_false_{negative,positive}_rate_at_thresholds. PiperOrigin-RevId: 172048554
* Internal change.Gravatar Anna R2017-10-12
| | | | PiperOrigin-RevId: 172045110
* Temporarily make tf.where emit an old-style NodeDef (cast non-bool inputs to ↵Gravatar Eugene Brevdo2017-10-12
| | | | | | bool) PiperOrigin-RevId: 172044654
* Determine peak memory usage from the result of a simulation.Gravatar Benoit Steiner2017-10-12
| | | | PiperOrigin-RevId: 172043591
* [tf.data] Actually add `tf.contrib.data.get_single_element()` to the public API.Gravatar Derek Murray2017-10-12
| | | | PiperOrigin-RevId: 172041381
* Automated g4 rollback of changelist 172019169Gravatar Anna R2017-10-12
| | | | PiperOrigin-RevId: 172041133
* Removes brittleness from some tests in dnn_test.Gravatar A. Unique TensorFlower2017-10-12
| | | | PiperOrigin-RevId: 172040631
* Switch to nest.flatten() in tf.layers.Layer to allow dicts andGravatar A. Unique TensorFlower2017-10-12
| | | | | | arbitrary nesting in layer inputs & outputs. PiperOrigin-RevId: 172040243
* TFE: Adds a destructor to ResourceVariables in Python that destroys the ↵Gravatar Ali Yahya2017-10-12
| | | | | | underlying resource. This makes the lifetime of the underlying resource match that of its corresponding Python object. PiperOrigin-RevId: 172039259
* [TF:XLA] Implement BitwiseAnd, BitwiseOr, and Invert operators.Gravatar A. Unique TensorFlower2017-10-12
| | | | PiperOrigin-RevId: 172038787
* typoGravatar A. Unique TensorFlower2017-10-12
| | | | PiperOrigin-RevId: 172037998
* BUILD cleanup in contrib/boosted_treesGravatar A. Unique TensorFlower2017-10-12
| | | | PiperOrigin-RevId: 172034428
* A few profiler improvements.Gravatar A. Unique TensorFlower2017-10-12
| | | | | | | | | 1. Use a id_to_string map to reduce the profile size (2/3 in xception) 2. dedup code view's function name with extra file base name. 3. remove code view display heuristic that doesn't work in some cases. 4. make the profile_context thread-safe. PiperOrigin-RevId: 172031528
* Puts the global step read within the global step name scope, for ↵Gravatar A. Unique TensorFlower2017-10-12
| | | | | | clean/organized graph displays in TensorBoard. PiperOrigin-RevId: 172028555
* Replace tf.conj(tf.transpose(foo)) and tf.transpose(tf.conj(foo)) idioms ↵Gravatar A. Unique TensorFlower2017-10-12
| | | | | | with tf.linalg.adjoint(foo) or tf.transpose(foo, conjugate=True), and clean up a few places that can avoid explicit adjoints as inputs to matmul. PiperOrigin-RevId: 172027859
* Make CheckNumerics on GPU asynchronous and remove the need for a hard device ↵Gravatar A. Unique TensorFlower2017-10-12
| | | | | | synchronization, which can have a significant performance impact. PiperOrigin-RevId: 172025744
* Internal changeGravatar Gunhan Gulsoy2017-10-12
| | | | PiperOrigin-RevId: 172023756
* Require that work_element_count be positive in GetCudaLaunchConfig() rather ↵Gravatar A. Unique TensorFlower2017-10-12
| | | | | | than trying to launch kernels with an invalid config and leaving the stream in an error state. The latter is much harder to debug. PiperOrigin-RevId: 172019169
* Add linux_arm64 and linux_armhf config_setting.Gravatar A. Unique TensorFlower2017-10-12
| | | | PiperOrigin-RevId: 172018709
* [XLA] Return status instead of TF_RET_CHECK.Gravatar Jacques Pienaar2017-10-12
| | | | | | | TF_RET_CHECK logs the call stack and HloEvaluator is often called with parameters and non-constant ops which results in a logging output in cases where the caller expects it could fail. PiperOrigin-RevId: 172018263
* Use canonical method for not changing the scope, but changing scope options.Gravatar A. Unique TensorFlower2017-10-12
| | | | PiperOrigin-RevId: 172014544
* Explictly raise an exception if sharding fails because the batch size is ↵Gravatar A. Unique TensorFlower2017-10-12
| | | | | | | | | | | unknown. Right now, TPUEstimator will raise an error message complaining that sharding failed because (None % integer value) is not a well-defined operation. I hope the new error message will make it easier for people to figure out what's going on. PiperOrigin-RevId: 172013663
* Update ops-related pbtxt files.Gravatar A. Unique TensorFlower2017-10-12
| | | | PiperOrigin-RevId: 172013619
* Internal change.Gravatar Toby Boyd2017-10-12
| | | | PiperOrigin-RevId: 172013289
* Multi-axis batch normalization support and changing from num_virtual_batches ↵Gravatar Chris Ying2017-10-12
| | | | | | | | | | | | to virtual_batch_size Key changes: 1) adding support for multi-axis batch norm by allowing the axis to be either an int or a list of ints 2) multi-axis batch norm is handled entirely by TensorFlow ops at the moment (no special kernel) and the performance is heavily dependent on the Tensor format (see reduce_* kernels for reduction rules) 3) Fix ghost batch norm by sharing the same gamma/beta/mean/var parameter across all virtual batches 4) Change ghost batch norm API to virtual_batch_size to be more consistent with its intended use case. PiperOrigin-RevId: 172012360
* Adding fused version of map and batch.Gravatar Jiri Simsa2017-10-12
| | | | PiperOrigin-RevId: 172009823
* Increase the size of extenders_test to avoid flaky timeouts.Gravatar Gunhan Gulsoy2017-10-12
| | | | PiperOrigin-RevId: 172005195
* Adds explicit type checks for all steps/iterations in TPUEstimator.Gravatar Jianwei Xie2017-10-12
| | | | PiperOrigin-RevId: 172001904
* Register a dummy estimation function for _Send ops. This ensures that we ↵Gravatar Benoit Steiner2017-10-12
| | | | | | don't get negative cost estimates for _Send ops. PiperOrigin-RevId: 171999586
* This creates a library under contrib for implementations of common metric ↵Gravatar A. Unique TensorFlower2017-10-12
| | | | | | | | | | | | | | learning losses, starting with: * [Contrastive loss](http://yann.lecun.com/exdb/publis/pdf/hadsell-chopra-lecun-06.pdf) * [Triplet Loss with semihard negative mining](https://arxiv.org/abs/1503.03832) * [Npairs loss](http://www.nec-labs.com/uploads/images/Department-Images/MediaAnalytics/papers/nips16_npairmetriclearning.pdf) * Npairs loss w/ multilabel support * [Lifted structured loss](http://www.cv-foundation.org/openaccess/content_cvpr_2016/papers/Song_Deep_Metric_Learning_CVPR_2016_paper.pdf) * [Structured clustering embedding](https://arxiv.org/abs/1612.01213) PiperOrigin-RevId: 171997156
* Java: Tweak to address some Javadoc errors.Gravatar Asim Shankar2017-10-12
| | | | PiperOrigin-RevId: 171987329
* Updated the virtual scheduler to use legal names when inserting Send/Recv ↵Gravatar Benoit Steiner2017-10-12
| | | | | | nodes in the graph. PiperOrigin-RevId: 171986401
* Change the default seed in TPUEstimator's RunConfig to be None.Gravatar Jonathan Hseu2017-10-12
| | | | PiperOrigin-RevId: 171983725
generated by cgit v1.2.3 (git 2.39.1) at 2024-07-26 15:08:38 +0000