tensorflow/models · error

Rank of Tensor %s must be known

Error message

Rank of Tensor %s must be known

What it means

Error "Rank of Tensor %s must be known" thrown in tensorflow/models.

Source

Thrown at official/projects/bigbird/recompute_grad.py:129

  """Force all of `then_compute` to depend on all of `first_compute`.

  Uses a dummy data dependency, which is useful when running on TPUs because
  XLA ignores control dependencies. Only supports float arguments.

  Args:
    first_compute: Sequence of `Tensor`s to be executed before `then_compute`.
    then_compute: Sequence of `Tensor`s to executed after `first_compute`.

  Returns:
    Sequence of `Tensor`s with same length of `then_compute`.

  Raises:
    ValueError: if ranks are unknown or types are not floating.
  """

  def _first_element(x):
    if x.shape.ndims is None:
      raise ValueError('Rank of Tensor %s must be known' % x)
    ndims = x.shape.ndims
    begin = tf.zeros(ndims, dtype=tf.int32)
    size = tf.ones(ndims, dtype=tf.int32)
    return tf.reshape(tf.slice(x, begin, size), [])

  first_compute_sum = tf.add_n(
      [_first_element(x) for x in first_compute if x is not None])
  dtype = first_compute_sum.dtype
  if not dtype.is_floating:
    raise ValueError('_force_data_dependency only supports floating dtypes.')
  zero = np.finfo(dtype.as_numpy_dtype).tiny * first_compute_sum
  return [  # pyrefly: ignore[bad-return]
      x + tf.cast(zero, x.dtype) if x is not None else None
      for x in then_compute
  ]


def _make_seed_if_none(seed: Optional[tf.Tensor]) -> tf.Tensor:

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/projects/bigbird/recompute_grad.py:129 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/719b583a9aab955d. Report an issue: GitHub.