tensorflow/models · error · ValueError

`num_classes` should be given when requesting one hot label.

Error message

`num_classes` should be given when requesting one hot label.

What it means

Error "`num_classes` should be given when requesting one hot label." thrown in tensorflow/models.

Source

Thrown at official/projects/video_ssl/dataloaders/video_ssl_input.py:199

    # together in batch dimenstion which will be 2 * batch.
    image = tf.concat(tf.split(image, num_or_size_splits=2, axis=1), axis=0)

  num_views = num_test_clips * num_test_crops
  if num_views > 1 and not is_training:
    # In this case, multiple views are merged together in batch dimenstion which
    # will be batch * num_views.
    image = tf.reshape(image, [-1, num_frames] + image.shape[2:].as_list())

  return image


def _process_label(label: tf.Tensor,
                   one_hot_label: bool = True,
                   num_classes: Optional[int] = None) -> tf.Tensor:
  """Processes label Tensor."""
  # Validate parameters.
  if one_hot_label and not num_classes:
    raise ValueError(
        '`num_classes` should be given when requesting one hot label.')

  # Cast to tf.int32.
  label = tf.cast(label, dtype=tf.int32)

  if one_hot_label:
    # Replace label index by one hot representation.
    label = tf.one_hot(label, num_classes)
    if len(label.shape.as_list()) > 1:
      label = tf.reduce_sum(label, axis=0)
    if num_classes == 1:
      # The trick for single label.
      label = 1 - label

  return label


class Parser(video_input.Parser):

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/projects/video_ssl/dataloaders/video_ssl_input.py:199 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/4793d23a361bcbe0. Report an issue: GitHub.