tensorflow/models · error · ValueError

Expect rank for tensor shape, but got None.

Error message

Expect rank for tensor shape, but got None.

What it means

Error "Expect rank for tensor shape, but got None." thrown in tensorflow/models.

Source

Thrown at official/nlp/modeling/ops/decoding_module.py:99

    shape_list_obj[-1] = None
  return tf.TensorShape(shape_list_obj)


def expand_to_same_rank(tensor, target):
  """Expands a given tensor to target's rank to be broadcastable.

  Args:
    tensor: input tensor to tile. Shape: [b, d1, ..., da]
    target: target tensor. Shape: [b, d1, ..., da, ..., dn]

  Returns:
    Tiled tensor of shape [b, d1, ..., da, 1, ..., 1] with same rank of target

  Raises:
    ValueError, if the shape rank of rank tensor/target is None.
  """
  if tensor.shape.rank is None:
    raise ValueError("Expect rank for tensor shape, but got None.")
  if target.shape.rank is None:
    raise ValueError("Expect rank for target shape, but got None.")

  with tf.name_scope("expand_rank"):
    diff_rank = target.shape.rank - tensor.shape.rank
    for _ in range(diff_rank):
      tensor = tf.expand_dims(tensor, -1)
    return tensor


class DecodingModule(tf.Module, metaclass=abc.ABCMeta):
  """A base class for the API required for decoding (go/decoding-tf-nlp)."""

  def __init__(self,
               length_normalization_fn: Callable[[int, tf.DType], float],
               dtype: tf.DType = tf.float32,
               decoding_name: Optional[str] = None,
               extra_cache_output: bool = False):

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/nlp/modeling/ops/decoding_module.py:99 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/65d78c367907049e. Report an issue: GitHub.