tensorflow/models · error · ValueError

Wrong shape detected for custom policy. Expected (:, :, 3) b

Error message

Wrong shape detected for custom policy. Expected (:, :, 3) but got {}.

What it means

Error "Wrong shape detected for custom policy. Expected (:, :, 3) but got {}." thrown in tensorflow/models.

Source

Thrown at official/vision/ops/augment.py:2006

      self.policies = self.available_policies[augmentation_name]

    else:
      self._check_policy_shape(policies)
      self.policies = policies

  def _check_policy_shape(self, policies):
    """Checks dimension and shape of the custom policy.

    Args:
      policies: List of list of tuples in the form `(func, prob, level)`. Must
        have shape of `(:, :, 3)`.

    Raises:
      ValueError if the shape of `policies` is unexpected.
    """
    in_shape = np.array(policies).shape
    if len(in_shape) != 3 or in_shape[-1:] != (3,):
      raise ValueError('Wrong shape detected for custom policy. Expected '
                       '(:, :, 3) but got {}.'.format(in_shape))

  def _make_tf_policies(self):
    """Prepares the TF functions for augmentations based on the policies."""
    replace_value = [128] * 3

    # func is the string name of the augmentation function, prob is the
    # probability of applying the operation and level is the parameter
    # associated with the tf op.

    # tf_policies are functions that take in an image and return an augmented
    # image.
    tf_policies = []
    for policy in self.policies:
      tf_policy = []
      assert_ranges = []
      # Link string name to the correct python function and make sure the
      # correct argument is passed into that function.

View on GitHub (pinned to e006f5f0d5)

Solutions

1. Shape the custom policy array as (:, :, 3): each sub-policy is a list of (op, prob, magnitude) triples.
2. Fix the custom policy structure so every operation entry has exactly 3 elements.

When it happens

Trigger: Thrown at official/vision/ops/augment.py:2006 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/db4698e65f312ac6. Report an issue: GitHub.