tensorflow/models · error · ValueError

examples_per_group={self._examples_per_group} is larger than

Error message

examples_per_group={self._examples_per_group} is larger than the number of examples available in the local (per-device) batch_size={batch_size}. Either decrease examples_per_group or increase the batch_size.

What it means

Error "examples_per_group={self._examples_per_group} is larger than the number of examples available in the local (per-device) batch_size={batch_size}. Either decrease examples_per_group or increase the batch_size." thrown in tensorflow/models.

Source

Thrown at official/nlp/modeling/layers/moe.py:588

        <float>[batch_size, seq_length, hidden_dim].
      training: Only apply dropout and jitter noise during training. If not
        provided taken from tf_keras.backend.

    Returns:
      Transformed inputs with same shape as inputs:
        <float>[batch_size, seq_length, hidden_dim].

    Raises:
      ValueError if we cannot find a group_size satisfying given requirements.
    """
    if training is None:
      training = tf_keras.backend.learning_phase()

    # inputs shape [batch_size, seq_length, hidden_dim]
    batch_size, seq_length, hidden_dim = inputs.shape
    if batch_size is not None:
      if self._examples_per_group > batch_size:
        raise ValueError(
            f"examples_per_group={self._examples_per_group} is larger than the "
            "number of examples available in the local (per-device) batch_size="
            f"{batch_size}. Either decrease examples_per_group or increase the "
            "batch_size.")
    tokens_per_group = int(seq_length * self._examples_per_group)

    if training:
      capacity_factor = self._train_capacity_factor
    else:
      capacity_factor = self._eval_capacity_factor
    # Each group will send expert_capacity tokens to each expert.
    expert_capacity = int(
        round(capacity_factor * tokens_per_group / self.num_experts))

    # Reshape batch and sequence/token dimensions for expert routing.
    x = tf.reshape(inputs, (-1, tokens_per_group, hidden_dim))

    x = self._mask_and_dispatch_to_experts(x, expert_capacity, training)

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/nlp/modeling/layers/moe.py:588 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/175e89a8c9a53130. Report an issue: GitHub.