tensorflow/models · error · TypeError

AverageModelCheckpoint is only used when trainingwith Moving

Error message

AverageModelCheckpoint is only used when trainingwith MovingAverage

What it means

Error "AverageModelCheckpoint is only used when trainingwith MovingAverage" thrown in tensorflow/models.

Source

Thrown at official/legacy/image_classification/callbacks.py:235

  """

  def __init__(self,
               update_weights: bool,
               filepath: str,
               monitor: str = 'val_loss',
               verbose: int = 0,
               save_best_only: bool = False,
               save_weights_only: bool = False,
               mode: str = 'auto',
               save_freq: str = 'epoch',
               **kwargs):
    self.update_weights = update_weights
    super().__init__(filepath, monitor, verbose, save_best_only,
                     save_weights_only, mode, save_freq, **kwargs)

  def set_model(self, model):
    if not isinstance(model.optimizer, optimization.ExponentialMovingAverage):
      raise TypeError('AverageModelCheckpoint is only used when training'
                      'with MovingAverage')
    return super().set_model(model)

  def _save_model(self, epoch, logs):
    assert isinstance(self.model.optimizer,
                      optimization.ExponentialMovingAverage)

    if self.update_weights:
      self.model.optimizer.assign_average_vars(self.model.variables)
      return super()._save_model(epoch, logs)  # pytype: disable=attribute-error  # typed-keras
    else:
      # Note: `model.get_weights()` gives us the weights (non-ref)
      # whereas `model.variables` returns references to the variables.
      non_avg_weights = self.model.get_weights()
      self.model.optimizer.assign_average_vars(self.model.variables)
      # result is currently None, since `super._save_model` doesn't
      # return anything, but this may change in the future.
      result = super()._save_model(epoch, logs)  # pytype: disable=attribute-error  # typed-keras

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/legacy/image_classification/callbacks.py:235 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/aae07f9899b672ba. Report an issue: GitHub.