tensorflow/models · error · ValueError

Unexpected dtype: %s

Error message

Unexpected dtype: %s

What it means

Error "Unexpected dtype: %s" thrown in tensorflow/models.

Source

Thrown at official/modeling/performance.py:53

      optimizer = tf_keras.mixed_precision.LossScaleOptimizer(
          optimizer, dynamic=False, initial_scale=loss_scale)
  return optimizer


def set_mixed_precision_policy(dtype, loss_scale=None):
  """Sets the global `tf_keras.mixed_precision.Policy`."""
  # TODO(b/191894773): Remove loss_scale argument
  assert loss_scale is None, (
      'The loss_scale argument must be None. The argument exists for '
      'historical reasons and will be removed soon.')
  if dtype == tf.float16:
    tf_keras.mixed_precision.set_global_policy('mixed_float16')
  elif dtype == tf.bfloat16:
    tf_keras.mixed_precision.set_global_policy('mixed_bfloat16')
  elif dtype == tf.float32:
    tf_keras.mixed_precision.set_global_policy('float32')
  else:
    raise ValueError('Unexpected dtype: %s' % dtype)

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/modeling/performance.py:53 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/1d279add19099a6b. Report an issue: GitHub.