tensorflow/models · error · ValueError

steps should be an Tensor. Python object may cause retracing

Error message

steps should be an Tensor. Python object may cause retracing.

What it means

Error "steps should be an Tensor. Python object may cause retracing." thrown in tensorflow/models.

Source

Thrown at official/legacy/xlnet/training_utils.py:202

      if input_meta_data["mem_len"] > 0:
        return mem

    def train_steps(iterator, steps):
      """Performs distributed training steps in a loop.

      Args:
        iterator: the distributed iterator of training datasets.
        steps: an tf.int32 integer tensor to specify number of steps to run
          inside host training loop.

      Raises:
        ValueError: Any of the arguments or tensor shapes are invalid.

      Returns:
        logits: logits computed.
      """
      if not isinstance(steps, tf.Tensor):
        raise ValueError("steps should be an Tensor. Python object may cause "
                         "retracing.")

      def cache_fn():
        """Initializes memory tensor used in XLNet pretraining."""
        mems = []
        if input_meta_data["mem_len"] > 0:
          for _ in range(input_meta_data["n_layer"]):
            zeros = tf.zeros([
                input_meta_data["batch_size_per_core"],
                input_meta_data["mem_len"],
                input_meta_data["d_model"]
            ],
                             dtype=tf.float32)
            mems.append(zeros)
        return mems

      if input_meta_data["mem_len"] > 0:
        mem = strategy.run(cache_fn)

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/legacy/xlnet/training_utils.py:202 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/6495f50019eb1901. Report an issue: GitHub.