754226949c99108ef22cbf8009e8c40efbbc7b5a,tf_agents/experimental/distributed/examples/sac/sac_train.py,,train,#Any#Any#Any#Any#Any#Any#Any#Any#Any#Any#,104

Before Change

  // Run the training loop.
  // TODO(b/162440911) change the loop use train_step to handle preemptions
  for _ in range(num_iterations):
    sac_learner.run(iterations=learner_iterations_per_call)
    variable_container.push(variables)


def main(_):
  logging.set_verbosity(logging.INFO)
  tf.enable_v2_behavior()

After Change

      strategy=strategy)

  // Run the training loop.
  while train_step.numpy() < num_iterations:
    sac_learner.run(iterations=learner_iterations_per_call)
    variable_container.push(variables)


def main(_):
  logging.set_verbosity(logging.INFO)
  tf.enable_v2_behavior()

Instances