tensorflow/models · error · ValueError

`train_input_fn`, `total_training_steps`, `steps_per_loop`,

Error message

`train_input_fn`, `total_training_steps`, `steps_per_loop`, `optimizer`, `save_steps` and `learning_rate_fn` are required parameters.

What it means

Error "`train_input_fn`, `total_training_steps`, `steps_per_loop`, `optimizer`, `save_steps` and `learning_rate_fn` are required parameters." thrown in tensorflow/models.

Source

Thrown at official/legacy/xlnet/training_utils.py:103

      init_from_transformerxl: Whether to load to `transformerxl_model` of
        `model_fn`.
      model_dir: The directory of model (checkpoints, summaries).
      save_steps: The frequency to save checkpoints. Every save_steps, we save a
        model checkpoint. Model checkpoint will be saved and evaluation will be
        conducted if evaluation dataset is provided.
      run_eagerly: Whether to run training eagerly.

  Returns:
      Last training step logits if training happens, otherwise returns None.
  Raises:
    TypeError: if model directory is not specified.
  """
  required_arguments = [
      train_input_fn, total_training_steps, steps_per_loop, optimizer,
      learning_rate_fn, save_steps
  ]
  if [arg for arg in required_arguments if arg is None]:
    raise ValueError("`train_input_fn`, `total_training_steps`, "
                     "`steps_per_loop`, `optimizer`, `save_steps` and "
                     "`learning_rate_fn` are required parameters.")
  if not model_dir:
    raise TypeError("Model directory must be specified.")
  train_iterator = data_utils.get_input_iterator(train_input_fn, strategy)
  if not tf.io.gfile.exists(model_dir):
    tf.io.gfile.mkdir(model_dir)
  # Create summary writers
  summary_dir = os.path.join(model_dir, "summaries")
  if not tf.io.gfile.exists(summary_dir):
    tf.io.gfile.mkdir(summary_dir)
  train_summary_writer = None
  eval_summary_writer = None
  if eval_fn:
    eval_summary_writer = tf.summary.create_file_writer(
        os.path.join(summary_dir, "eval"))
  if steps_per_loop >= _MIN_SUMMARY_STEPS:
    # Only writes summary when the stats are collected sufficiently over

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/legacy/xlnet/training_utils.py:103 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/9982bd6f50f5bbe8. Report an issue: GitHub.