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

Before Change

    replay_buffer = tf_uniform_replay_buffer.TFUniformReplayBuffer(
        spec, batch_size=batch_size)

    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())

After Change

          num_steps=2, time_stacked=False)

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

    for _ in range(100):
      (step_, next_step_), _ = self.evaluate(sample)
      self.assertEqual((step_ + 1) % 10, next_step_)

Instances