tensorflow/models · error · ValueError

`model` must be provided if using `ExponentialMovingAverage`

Error message

`model` must be provided if using `ExponentialMovingAverage`.

What it means

Error "`model` must be provided if using `ExponentialMovingAverage`." thrown in tensorflow/models.

Source

Thrown at official/legacy/image_classification/optimizer_factory.py:270

    optimizer = legacy_adamw.AdamWeightDecay(
        learning_rate=base_learning_rate,
        weight_decay_rate=weight_decay,
        beta_1=beta_1,
        beta_2=beta_2,
        epsilon=epsilon,
    )
  else:
    raise ValueError('Unknown optimizer %s' % optimizer_name)

  if params.get('lookahead', None):
    logging.info('Using lookahead optimizer.')
    optimizer = Lookahead(optimizer)

  # Moving average should be applied last, as it's applied at test time
  moving_average_decay = params.get('moving_average_decay', 0.)
  if moving_average_decay is not None and moving_average_decay > 0.:
    if model is None:
      raise ValueError(
          '`model` must be provided if using `ExponentialMovingAverage`.')
    logging.info('Including moving average decay.')
    optimizer = optimization.ExponentialMovingAverage(
        optimizer=optimizer, average_decay=moving_average_decay)
    optimizer.shadow_copy(model)
  return optimizer


def build_learning_rate(params: base_configs.LearningRateConfig,
                        batch_size: Optional[int] = None,
                        train_epochs: Optional[int] = None,
                        train_steps: Optional[int] = None):
  """Build the learning rate given the provided configuration."""
  decay_type = params.name
  base_lr = params.initial_lr
  decay_rate = params.decay_rate
  if params.decay_epochs is not None:
    decay_steps = params.decay_epochs * train_steps

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/legacy/image_classification/optimizer_factory.py:270 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of tensorflow/models@e006f5f0d5 (2026-08-24). Data as JSON: /api/errors/b9baed1e28208720. Report an issue: GitHub.