HomeSort by relevance Sort by last modified time
    Searched refs:per_host_input_for_training (Results 1 - 4 of 4) sorted by null

  /external/tensorflow/tensorflow/python/tpu/
tpu_config.py 52 'per_host_input_for_training',
75 per_host_input_for_training: If `True`, `PER_HOST_V1`, or `PER_HOST_V2`,
86 Note: per_host_input_for_training==PER_SHARD_V1 only supports mode.TRAIN.
107 replicas. Unlike per_host_input_for_training=BROADCAST, each replica will
109 the behaviour is determined by per_host_input_for_training.
120 per_host_input_for_training=True,
140 if per_host_input_for_training is not InputPipelineConfig.PER_HOST_V2:
162 # per_host_input_for_training may be True, False, or integer in [1..3].
164 if per_host_input_for_training is False:
165 per_host_input_for_training = InputPipelineConfig.PER_SHARD_V
    [all...]
tpu_context.py 60 The configuration depends on `TPUConfig.per_host_input_for_training`. See
398 (self._config.tpu_config.per_host_input_for_training is
403 return (self._config.tpu_config.per_host_input_for_training is
409 return ((self._config.tpu_config.per_host_input_for_training is
tpu_estimator.py 752 'TPUConfig.per_host_input_for_training to True or return '
    [all...]
  /external/tensorflow/tensorflow/contrib/eager/python/examples/revnet/
main_estimator_tpu.py 205 per_host_input_for_training=tf.contrib.tpu.InputPipelineConfig.

Completed in 3692 milliseconds