tensorflow/models · error · ValueError

`steps_per_loop` is required when `trainer` is provided.

Error message

`steps_per_loop` is required when `trainer` is provided.

What it means

Error "`steps_per_loop` is required when `trainer` is provided." thrown in tensorflow/models.

Source

Thrown at orbit/controller.py:187

        created internally for TensorBoard summaries by default from the
        `summary_dir`.
      eval_summary_manager: Instance of the eval summary manager. If set, the
        `eval_summary_dir` will be ignored. Otherwise the eval summary manager
        will be created internally for TensorBoard summaries by default from the
        `eval_summary_dir`.

    Raises:
      ValueError: If both `trainer` and `evaluator` are `None`.
      ValueError: If `steps_per_loop` is not a positive integer or a callable.
      ValueError: If `summary_interval` is not a positive integer or is not
        divisible by `steps_per_loop`.
    """
    if trainer is None and evaluator is None:
      raise ValueError("`trainer` and `evaluator` should not both be `None`.")

    if trainer is not None:
      if steps_per_loop is None:
        raise ValueError(
            "`steps_per_loop` is required when `trainer` is provided.")
      elif not callable(steps_per_loop) and (
          not isinstance(steps_per_loop, int) or steps_per_loop < 1):
        raise ValueError(
            f"`steps_per_loop` ({steps_per_loop}) must be a positive integer "
            "or a callable.")

      if summary_interval is not None:
        if summary_interval <= 0:
          raise ValueError(
              f"`summary_interval` ({summary_interval}) must be larger than 0.")
        elif not callable(steps_per_loop) and (summary_interval % steps_per_loop
                                               != 0):
          raise ValueError(
              f"`summary interval` ({summary_interval}) must be a multiple "
              f"of `steps_per_loop` ({steps_per_loop}).")

    if not isinstance(global_step, tf.Variable):

View on GitHub (pinned to e006f5f0d5)

Solutions

1. Pass steps_per_loop when constructing the Controller with a trainer.
2. Set steps_per_loop to the number of steps executed per train loop iteration.

When it happens

Trigger: Thrown at orbit/controller.py:187 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/3768f9726bd9b9e0. Report an issue: GitHub.