Missed TensorFlow Dev Summit? Check out the video playlist. Watch recordings

tf.distribute.RunOptions

View source on GitHub

Run options for strategy.run.

@staticmethod
tf.distribute.RunOptions(
    cls, experimental_enable_dynamic_batch_size=True,
    experimental_bucketizing_dynamic_shape=False
)

This can be used to hold some strategy specific configs.

Attributes:

  • experimental_enable_dynamic_batch_size: Boolean. Only applies to TPUStrategy. Default to True. If True, TPUStrategy will enable dynamic padder to support dynamic batch size for the inputs. Otherwise only static shape inputs are allowed.
  • experimental_bucketizing_dynamic_shape: Boolean. Only applies to TPUStrategy. Default to False. If True, TPUStrategy will automatic bucketize inputs passed into run if the input shape is dynamic. This is a performance optimization to reduce XLA recompilation, which should not have impact on correctness.