Commits

Branch selector

 master

User selector

jsimsa

Datepicker

All time

Commit History

Commits on Dec 22, 2018

• [tf.data] Minor refactoring of the parallel interleave kernel.

  Show description for 0dd93b6
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  0dd93b6

  Copy full SHA for 0dd93b6

Commits on Dec 20, 2018

• [tf.data] Making it possible to override modeling framework implementations.

  Show description for 10a114c
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  10a114c

  Copy full SHA for 10a114c

• Adding `Name` to the `RandomAccessFile` and `WritableFile` API.

  Show description for 4ccb03f
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  4ccb03f

  Copy full SHA for 4ccb03f

• [tf.keras] Adding utility methods for analyzing whether a dataset is not batched / shuffled.

  Show description for 110a4fd
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  110a4fd

  Copy full SHA for 110a4fd

• [tf.data] Refining the modeling framework to take buffer size into account.

  Show description for b6185de
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  b6185de

  Copy full SHA for b6185de

Commits on Dec 14, 2018

• [tf.data] Create resource manager for `ReduceDatasetOp` and `ToSingleElementOp`.

  Show description for c406a36
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  c406a36

  Copy full SHA for c406a36

Commits on Dec 13, 2018

• Adding deprecation warnings for I/O modules that will be removed in 2.0.

  Show description for 5d2c0ee
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  5d2c0ee

  Copy full SHA for 5d2c0ee

• [tf.data] Reduce the overhead of performance modeling when there are no autotunable knobs.

  Show description for fb83aa5
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  fb83aa5

  Copy full SHA for fb83aa5

Commits on Dec 8, 2018

• [tf.data] Replace internal mechanism for copying batch elements in MapAndBatchDataset as the current mechanism could result in a segfault if the elements to be batched are DT_VARIANT.

  Show description for 323f3eb
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  323f3eb

  Copy full SHA for 323f3eb

Commits on Dec 6, 2018

• [tf.data] Add documentation for supporting `tf.data.experimental.AUTOTUNE` as a value for the `num_parallel_calls` argument of `map()`, `interleave()`, and `map_and_batch()`.

  Show description for 2947a28
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  2947a28

  Copy full SHA for 2947a28

• [tf.data] Adding support for tracking allocated memory of tensors stored in internal buffers.

  Show description for 55e470a
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  55e470a

  Copy full SHA for 55e470a

Commits on Dec 5, 2018

• [tf.data] Changing semantics of the map family of datasets to be cardinality preserving (in the presence of OutOfRange errors produced by the user-defined function).

  Show description for d0bae80
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  d0bae80

  Copy full SHA for d0bae80

Commits on Dec 4, 2018

• [tf.data] Adding `tf.data.experimental.cardinality()` which provides information about dataset cardinality.

  Show description for 98eb7d8
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  98eb7d8

  Copy full SHA for 98eb7d8

• [tf.data] Moving the experimental RandomDataset and ScanDataset C++ kernels to `experimental/` and removing both the Python and C++ implementation of EnqueueInQueueDataset and PrependFromQueueAndPa…

  Show description for 1221a9f
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  1221a9f

  Copy full SHA for 1221a9f

Commits on Dec 2, 2018

• [tf.data] Moving the "visibility: HIDDEN" API definition file attribute from Python API (overrides) to the base API.

  Show description for 823cf73
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  823cf73

  Copy full SHA for 823cf73

• Fix API definition filename for TensorListPushBackBatch.

  Show description for e4e9391
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  e4e9391

  Copy full SHA for e4e9391

• [tf.data] Consolidating experimental C++ kernels.

  Show description for bf70c44
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  bf70c44

  Copy full SHA for bf70c44

Commits on Nov 29, 2018

• [tf.data] Nesting `tf.data.Options()` optimization options under `experimental_optimization`.

  Show description for 84f93b8
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  84f93b8

  Copy full SHA for 84f93b8

• [tf.data] Mitigating the effect map transformation stragglers have on the performance of the MapAndBatchDataset kernel.

  Show description for c25a017
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  c25a017

  Copy full SHA for c25a017

Commits on Nov 26, 2018

• [tf.data] Test and benchmark cleanup.

  Show description for 04e95d6
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  04e95d6

  Copy full SHA for 04e95d6

• [tf.data] Fix serialization bug in PrivateThreadPoolDatasetOp.

  Show description for e468989
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  e468989

  Copy full SHA for e468989

• [tf.data] Fix threadpool names.

  Show description for a70c22d
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  a70c22d

  Copy full SHA for a70c22d

Commits on Nov 22, 2018

• [tf.data] Moving the (experimental) MatchingFiles dataset implementation to the experimental source subtree.

  Show description for 1e4e632
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  1e4e632

  Copy full SHA for 1e4e632

Commits on Nov 21, 2018

• [tf.data] Options-related changes.

  Show description for fa5590b
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  fa5590b

  Copy full SHA for fa5590b

• [tf.data] Properly guard access to shared state with a mutex.

  Show description for 1f9055c
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  1f9055c

  Copy full SHA for 1f9055c

Commits on Nov 20, 2018

• [tf.data] Propagate private threadpool size parameter during iterator creation so that it is (actually) used as the maximum degree of parallelism for autotuning.

  Show description for 97a16c7
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  97a16c7

  Copy full SHA for 97a16c7

Commits on Nov 16, 2018

• [tf.data] Fix the default thread pool size use for auto-tuning.

  Show description for 2c1edcb
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  2c1edcb

  Copy full SHA for 2c1edcb

Commits on Nov 13, 2018

• [tf.data] Failing early when batch memory allocation fails.

  Show description for b4cc45a
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  b4cc45a

  Copy full SHA for b4cc45a

Commits on Nov 12, 2018

• [tf.data] Reducing the overhead of performance modeling.

  Show description for b88372e
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  b88372e

  Copy full SHA for b88372e

Commits on Nov 5, 2018

• [tf.data] Fix internal test failure.

  Show description for ae24dac
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  ae24dac

  Copy full SHA for ae24dac

• [tf.data] Fix internal test failure.

  Show description for 9e93f7f
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  9e93f7f

  Copy full SHA for 9e93f7f

• [tf.data] Clean up of how `IteratorContext` is constructed and modified.

  Show description for 059dd69
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  059dd69

  Copy full SHA for 059dd69

Commits on Nov 2, 2018

• [tf.data] Making it possible to apply static optimizations to input pipelines that contain datasets that do not implement serialization.

  Show description for e359bde
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  e359bde

  Copy full SHA for e359bde

• [tf.data] Enable performance modeling by default.

  Show description for b53c602
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  b53c602

  Copy full SHA for b53c602

• [tf.data] Fix an internal test failure.

  Show description for 1862911
  

  jsimsa

  authored and

  tensorflower-gardener

  committed
  1862911

  Copy full SHA for 1862911

Pagination

Previous

Next