tensorflow/models · error · ValueError

`dataset_or_fn` should be either callable or an instance of

Error message

`dataset_or_fn` should be either callable or an instance of `tf.data.Dataset`.

What it means

Error "`dataset_or_fn` should be either callable or an instance of `tf.data.Dataset`." thrown in tensorflow/models.

Source

Thrown at orbit/utils/common.py:74

    *args: Any positional arguments to pass through to `dataset_or_fn`.
    **kwargs: Any keyword arguments to pass through to `dataset_or_fn`, except
      that the `input_options` keyword is used to specify a
      `tf.distribute.InputOptions` for making the distributed dataset.

  Returns:
    A distributed Dataset.
  """
  if strategy is None:
    strategy = tf.distribute.get_strategy()

  input_options = kwargs.pop("input_options", None)

  if isinstance(dataset_or_fn, tf.data.Dataset):
    return strategy.experimental_distribute_dataset(dataset_or_fn,
                                                    input_options)

  if not callable(dataset_or_fn):
    raise ValueError("`dataset_or_fn` should be either callable or an instance "
                     "of `tf.data.Dataset`.")

  def dataset_fn(input_context):
    """Wraps `dataset_or_fn` for strategy.distribute_datasets_from_function."""

    # If `dataset_or_fn` is a function and has an argument named
    # `input_context`, pass through the given `input_context`. Otherwise
    # `input_context` will be ignored.
    argspec = inspect.getfullargspec(dataset_or_fn)
    arg_names = argspec.args

    if "input_context" in arg_names:
      kwargs["input_context"] = input_context
    return dataset_or_fn(*args, **kwargs)

  return strategy.distribute_datasets_from_function(dataset_fn, input_options)

View on GitHub (pinned to e006f5f0d5)

Solutions

1. Pass a tf.data.Dataset instance or a callable returning one.
2. Wrap dataset creation in a function if you need lazy construction.

When it happens

Trigger: Thrown at orbit/utils/common.py:74 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/ad0d4e6aa639e3a4. Report an issue: GitHub.