f01418596956fdd328dfa2d5795093c3fb6bb172,official/modeling/training/distributed_executor.py,DistributedExecutor,train,#DistributedExecutor#Any#Any#Any#Any#Any#Any#Any#Any#Any#Any#Any#,382

Before Change

        train_metric.reset_states()

    // Reaches the end of training and saves the last checkpoint.
    _save_checkpoint(checkpoint, model_dir,
                     checkpoint_name.format(step=current_step))

    if test_step:
      logging.info("Running final evaluation after training is complete.")
      eval_iterator = self._get_input_iterator(eval_input_fn, strategy)

After Change

        train_metric.reset_states()

    // Reaches the end of training and saves the last checkpoint.
    if last_save_checkpoint_step < total_steps:
      _save_checkpoint(checkpoint, model_dir,
                       checkpoint_name.format(step=current_step))

    if test_step:
      logging.info("Running final evaluation after training is complete.")
      eval_iterator = self._get_input_iterator(eval_input_fn, strategy)
      eval_metric_result = self._run_evaluation(test_step, current_step,

Instances