Permalink
Commits on Mar 3, 2019
Commits on Mar 2, 2019
-
… to zero out trailing row or column. PiperOrigin-RevId: 236483699
-
…puts. Re-landing after this was rolled back. PiperOrigin-RevId: 236483500
-
Previously it was only tested under eager mode, and new issue was discovered by when "TF2_BEHAVIOR=1". Also the v2 tests has been updated to use keras.LSTMCell to ensure the correctness in v2. PiperOrigin-RevId: 236470595
-
-
This change is a prerequisite for introducing C Ops and Kernels into Android and iOS builds. PiperOrigin-RevId: 236432247
-
… of the pool created when multiprocessing=True is set. PiperOrigin-RevId: 236417625
-
The previous implementation of many core `tf.data` transformations (e.g. `Dataset.prefetch()`) would create one or more threads each time an iterator over those datasets is created (e.g. `ds.prefetch(N).repeat(100)` would create and destroy 100 threads). In addition to the overhead of thread creation, this interacts poorly with some malloc implementations, and can contribute to memory fragmentation. The new implementation maintains an unbounded pool of physical threads in each iterator (or `MultiDeviceIterator`) resource, and returns logical "threads" to that pool when their work is complete instead of exiting from them. PiperOrigin-RevId: 236413014
-
Before: https://www.tensorflow.org/versions/r2.0/api_docs/python/tf/optimizers/Adagrad PiperOrigin-RevId: 236410591
-
When calling Keras add_weight with a np list, as written the `shape or ()` "trick" results in the following exception: """ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()""" This change fixes the problem by using an explicit `if`. PiperOrigin-RevId: 236407103
-
PiperOrigin-RevId: 236405963
-
Almost all of the logic of the CopyRemover class was contained in an inner class, BufferValueTracker. This change moves the logic that was inside of CopyRemover into callers, and promotes BufferValueTracker into a new CopyRemover class. PiperOrigin-RevId: 236404892
-
…es to make them consistent. PiperOrigin-RevId: 236400962
Commits on Mar 1, 2019
-
TensorShape.is_compatible_with is symmetric; we need an asymmetric test here. PiperOrigin-RevId: 236390944
-
Imported from GitHub PR #21207 if one includes the header and does not link with coding.cc gets unresolved error. not the case when you just include extern functions and do not use them, more over this function does not seem to save so much being inlined. Copybara import of the project: - 60ab66e removed inline function by Stefan Dyulgerov <stefan.dyulgerov@gmail.com> - ba7b6ed Merge 60ab66e into 132d1... by Stefan Dyulgerov <stefan.dyulgerov@gmail.com> COPYBARA_INTEGRATE_REVIEW=#21207 from kingofthebongo2008:clean_coding_refactor 60ab66e PiperOrigin-RevId: 236390630
NewerOlder