060c9f88fc890bce5657fdf2241e910e9291438b,tf_agents/bandits/agents/greedy_reward_prediction_agent_test.py,,_get_initial_and_final_steps,#Any#Any#,65

Before Change

          name="step_type"),
      tf.constant(rewards, dtype=tf.float32, name="reward"),
      tf.constant(1.0, dtype=tf.float32, shape=[batch_size], name="discount"),
      tf.constant(observations + 100.0, dtype=tf.float32, name="observation"))
  return initial_step, final_step

After Change

def _get_initial_and_final_steps(observations, rewards):
  batch_size = tf.nest.flatten(observations)[0].shape[0]
  if isinstance(observations, np.ndarray):
    observations = tf.constant(
        observations, dtype=tf.float32, name="observation")
  initial_step = ts.TimeStep(
      tf.constant(
          ts.StepType.FIRST,
          dtype=tf.int32,

Instances