c611d9b6b35e44326bc18096dc57abe18eb6c90b,tf_agents/replay_buffers/tf_uniform_replay_buffer_test.py,TFUniformReplayBufferTest,testMultiStepSampling,#TFUniformReplayBufferTest#Any#,208

Before Change

    action = tf.stack([tf.Variable(0).count_up_to(10)] * batch_size)

    add_op = replay_buffer.add_batch(action)
    (step, next_step), _ = replay_buffer.get_next(num_steps=2,
                                                  time_stacked=False)

    self.evaluate(tf.compat.v1.global_variables_initializer())
    for _ in range(10):
      self.evaluate(add_op)

After Change

    if tf.executing_eagerly():
      sample = lambda: replay_buffer.get_next(num_steps=2, time_stacked=False)
    else:
      sample = replay_buffer.get_next(
          num_steps=2, time_stacked=False)

    self.evaluate(tf.compat.v1.global_variables_initializer())
    self.evaluate(add_data())

Instances