Change distribution.distribute_dataset to accept an input_fn instead of a dataset.
authorYuefeng Zhou <yuefengz@google.com>
Wed, 18 Apr 2018 23:35:44 +0000 (16:35 -0700)
committerTensorFlower Gardener <gardener@tensorflow.org>
Wed, 18 Apr 2018 23:38:07 +0000 (16:38 -0700)
commitfddfa9f8dcd1a922ade5362c0538ca39e99472a7
tree7320668c20041c85bcce4e0c6614f7270dce44c6
parente9d47fbff0d644a75c6f3dcdcb852685ef515b64
Change distribution.distribute_dataset to accept an input_fn instead of a dataset.

PiperOrigin-RevId: 193437651
tensorflow/contrib/distribute/python/minimize_loss_test.py
tensorflow/contrib/distribute/python/mirrored_strategy.py
tensorflow/contrib/distribute/python/mirrored_strategy_multigpu_test.py
tensorflow/contrib/distribute/python/one_device_strategy.py
tensorflow/contrib/distribute/python/optimizer_v2_test.py
tensorflow/contrib/distribute/python/single_loss_example.py
tensorflow/contrib/distribute/python/step_fn.py
tensorflow/python/BUILD
tensorflow/python/estimator/estimator.py
tensorflow/python/training/distribute.py