tensorflow/models · error · ValueError

`window_size` must be positive.

Error message

`window_size` must be positive.

What it means

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

Source

Thrown at official/projects/triviaqa/inputs.py:360

  def _make_relative_id_pattern(
      self, window_size: Union[int, tf.Tensor]) -> tf.Tensor:
    """Helper for making the relative id pattern for a particular window size.

    For example, if `max_distance` is 3, `ignore_direction` is False, and
    `window_size` is 11, the result is the following:
    [6, 6, 6, 5, 4, 0, 1, 2, 3, 3, 3].

    Args:
      window_size: Window size to return relative ids for. Must be positive and
        odd since ids will be relative to the center of the window. If a Tensor,
        must be a scalar int.

    Returns:
      <int32>[window_size] Tensor of relative position ids.
    """
    if isinstance(window_size, int):
      if window_size < 1:
        raise ValueError('`window_size` must be positive.')
      if window_size % 2 != 1:
        raise ValueError('`window_size` must be odd.')

    x = tf.range(self.max_distance + 1, dtype=tf.int32)
    x = tf.pad(x, [[self.max_distance, 0]], mode='REFLECT')
    if not self.ignore_direction:
      direction_adder = tf.concat([
          tf.fill([self.max_distance], self.max_distance),
          tf.zeros([self.max_distance + 1], dtype=tf.int32)
      ], 0)
      x += direction_adder

    len_x = x.shape.as_list()[0]
    if len_x > window_size:
      trim_amount = (len_x - window_size) // 2
      return x[trim_amount:-trim_amount]

    pad_amount = (window_size - len_x) // 2

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/projects/triviaqa/inputs.py:360 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/cc5cc5e2760294ce. Report an issue: GitHub.