tensorflow/models · error · ValueError

`seq_len` must be positive.

Error message

`seq_len` must be positive.

What it means

Error "`seq_len` must be positive." thrown in tensorflow/models.

Source

Thrown at official/projects/triviaqa/inputs.py:269

          [6, 5, 4, 0, 1, 2],
          [6, 6, 5, 4, 0, 1],
          [6, 6, 6, 5, 4, 0],
      ]]

    Args:
      seq_len: The sequence length to create ids for. Must be positive. If a
        Tensor, must be a scalar int.
      batch_size: The batch size of the result (default 1). Must be positive. If
        a Tensor, must be a scalar int. All examples in the batch will have the
        same id pattern.
      name: A name for the operation (optional).

    Returns:
      <int32>[batch_size, seq_len, seq_len] Tensor of relative position ids.
    """
    with tf.name_scope(name or 'make_relative_att_ids'):
      if isinstance(seq_len, int) and seq_len < 1:
        raise ValueError('`seq_len` must be positive.')
      if isinstance(batch_size, int) and batch_size < 1:
        raise ValueError('`batch_size` must be positive.')

      # We need the id_pattern to cover all tokens to the left of the last token
      # and all tokens to the right of the first token at the same time.
      window_size = 2 * seq_len - 1

      # [window_size]
      id_pattern = self._make_relative_id_pattern(window_size)

      # [seq_len, window_size]
      id_tensor = tf.tile(id_pattern[tf.newaxis, :], [seq_len, 1])

      # [seq_len, window_size + seq_len - 1]
      id_tensor = _skew_elements_right(id_tensor, -1)

      # [seq_len, seq_len]
      id_tensor = tf.slice(id_tensor, [0, seq_len - 1], [seq_len, seq_len])

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/projects/triviaqa/inputs.py:269 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/5eb062dc1153ddd3. Report an issue: GitHub.