metric_specs.py 文件源码

def create_metric_ops(self, _inputs, labels, predictions):
    """Creates (value, update_op) tensors
    """
    with tf.variable_scope(self._name):

      # Join tokens into single strings
      predictions_flat = tf.reduce_join(
          predictions["predicted_tokens"], 1, separator=self._separator)
      labels_flat = tf.reduce_join(
          labels["target_tokens"], 1, separator=self._separator)

      sources_value, sources_update = accumulate_strings(
          values=predictions_flat, name="sources")
      targets_value, targets_update = accumulate_strings(
          values=labels_flat, name="targets")

      metric_value = tf.py_func(
          func=self._py_func,
          inp=[sources_value, targets_value],
          Tout=tf.float32,
          name="value")

    with tf.control_dependencies([sources_update, targets_update]):
      update_op = tf.identity(metric_value, name="update_op")

    return metric_value, update_op