5ef64b79a43ba8f4806d68fe61d6a97ce25e6f80,models/official/mnasnet/mnasnet_main.py,,mnasnet_model_fn,#Any#Any#Any#Any#,262

Before Change

  if has_moving_average_decay:
    ema = tf.train.ExponentialMovingAverage(
        decay=params["moving_average_decay"], num_updates=global_step)
    ema_vars = tf.trainable_variables() + tf.get_collection("moving_vars")
    for v in tf.global_variables():
      // We maintain mva for batch norm moving mean and variance as well.
      if "moving_mean" in v.name or "moving_variance" in v.name:
        ema_vars.append(v)
    ema_vars = list(set(ema_vars))

  host_call = None
  if is_training:
    // Compute the current epoch and associated learning rate from global_step.

After Change

  if has_moving_average_decay:
    ema = tf.train.ExponentialMovingAverage(
        decay=params["moving_average_decay"], num_updates=global_step)
    ema_vars = mnasnet_utils.get_ema_vars()

  host_call = None
  if is_training:
    // Compute the current epoch and associated learning rate from global_step.

Instances