correct_prediction = tf.equal(tf.argmax(logits, 1), tf.argmax(labels, 1))
    accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
    optimizer.apply_gradients(zip(grads, model.trainable_variables))
    return accuracy, loss_value

  @tf.function
  def distributed_train_step(dist_inputs):
    per_replica_accuracy, per_replica_losses = strategy.run(

## Known Caveats

*   `tf.keras.mixed_precision`
    *   When using mixed precision, calling `RMSprop.apply_gradients` or
        `Nadam.apply_gradients` outside a `tf.function` does not work and will
        raise the AttributeError "Tensor.op is meaningless when eager execution
        is enabled". See this