Jay Shi 655dde4d59 [tf.data] Add the optimization to enable batch parallelization. ...

PiperOrigin-RevId: 356384660
Change-Id: I00024090fefa9410ab2da9ea62bca71f2934b8da

2021-02-08 17:08:02 -08:00

..

benchmarks

[tf.data] Implements the parallel version of BatchDataset. The dataset will do the copying elements work of BatchDataset parallelly. The CL also updates the modeling of the AsyncKnownRatio node to correctly model the dataset.

2021-02-01 17:01:49 -08:00

experimental

[tf.data] Add the optimization to enable batch parallelization.

2021-02-08 17:08:02 -08:00

kernel_tests

[tf.data] Default autotuning to conservative values to avoid accidentally allocating too much memory before optimization loop picks values that respect the memory budget.

2021-02-05 16:12:33 -08:00

ops

[tf.data] Add a new proto to store tf.data.Options data and conversion functions between tf.data.Options and their proto representation.

2021-02-05 15:55:03 -08:00

util

[tf.data] Add a new proto to store tf.data.Options data and conversion functions between tf.data.Options and their proto representation.

2021-02-05 15:55:03 -08:00

__init__.py

[tf.data] Graduating tf.data.experimental.AUTOTUNE to core API.

2020-09-09 19:08:06 -07:00

BUILD

PY2 removal cleanup

2021-01-15 16:48:57 -08:00