Function create_optimizer

a00_Bert/optimization.py:25–81

Creates an optimizer training op.

(loss, init_lr, num_train_steps, num_warmup_steps, use_tpu)

def create_optimizer(loss, init_lr, num_train_steps, num_warmup_steps, use_tpu):
  """Creates an optimizer training op."""
  global_step = tf.train.get_or_create_global_step()

  learning_rate = tf.constant(value=init_lr, shape=[], dtype=tf.float32)

  # Implements linear decay of the learning rate.
  learning_rate = tf.train.polynomial_decay(
      learning_rate,
      global_step,
      num_train_steps,
      end_learning_rate=0.0,
      power=1.0,
      cycle=False)

  # Implements linear warmup. I.e., if global_step < num_warmup_steps, the
  # learning rate will be `global_step/num_warmup_steps * init_lr`.
  if num_warmup_steps:
    global_steps_int = tf.cast(global_step, tf.int32)
    warmup_steps_int = tf.constant(num_warmup_steps, dtype=tf.int32)

    global_steps_float = tf.cast(global_steps_int, tf.float32)
    warmup_steps_float = tf.cast(warmup_steps_int, tf.float32)

    warmup_percent_done = global_steps_float / warmup_steps_float
    warmup_learning_rate = init_lr * warmup_percent_done

    is_warmup = tf.cast(global_steps_int < warmup_steps_int, tf.float32)
    learning_rate = (
        (1.0 - is_warmup) * learning_rate + is_warmup * warmup_learning_rate)

  # It is recommended that you use this optimizer for fine tuning, since this
  # is how the model was trained (note that the Adam m/v variables are NOT
  # loaded from init_checkpoint.)
  optimizer = AdamWeightDecayOptimizer(
      learning_rate=learning_rate,
      weight_decay_rate=0.01,
      beta_1=0.9,
      beta_2=0.999,
      epsilon=1e-6,
      exclude_from_weight_decay=["LayerNorm", "layer_norm", "bias"])

  if use_tpu:
    optimizer = tf.contrib.tpu.CrossShardOptimizer(optimizer)

  tvars = tf.trainable_variables()
  grads = tf.gradients(loss, tvars)

  # This is how the model was pre-trained.
  (grads, _) = tf.clip_by_global_norm(grads, clip_norm=1.0)

  train_op = optimizer.apply_gradients(
      zip(grads, tvars), global_step=global_step)

  new_global_step = global_step + 1
  train_op = tf.group(train_op, [global_step.assign(new_global_step)])
  return train_op