tf.contrib.training.train

tf.contrib.training.train(
    train_op,
    logdir,
    master='',
    is_chief=True,
    scaffold=None,
    hooks=None,
    chief_only_hooks=None,
    save_checkpoint_secs=600,
    save_summaries_steps=100,
    config=None,
    max_wait_secs=7200
)

Defined in tensorflow/contrib/training/python/training/training.py.

Runs the training loop.

Args:

  • train_op: A Tensor that, when executed, will apply the gradients and return the loss value.
  • logdir: The directory where the graph and checkpoints are saved.
  • master: The URL of the master.
  • is_chief: Specifies whether or not the training is being run by the primary replica during replica training.
  • scaffold: An tf.train.Scaffold instance.
  • hooks: List of tf.train.SessionRunHook callbacks which are run inside the training loop.
  • chief_only_hooks: List of tf.train.SessionRunHook instances which are run inside the training loop for the chief trainer only.
  • save_checkpoint_secs: The frequency, in seconds, that a checkpoint is saved using a default checkpoint saver. If save_checkpoint_secs is set to None, then the default checkpoint saver isn't used.
  • save_summaries_steps: The frequency, in number of global steps, that the summaries are written to disk using a default summary saver. If save_summaries_steps is set to None, then the default summary saver isn't used.
  • config: An instance of tf.ConfigProto.
  • max_wait_secs: Maximum time workers should wait for the session to become available. This should be kept relatively short to help detect incorrect code, but sometimes may need to be increased if the chief takes a while to start up.

Returns:

the value of the loss function after training.

Raises:

  • ValueError: if logdir is None and either save_checkpoint_secs or save_summaries_steps are `None.