Animation for the DQN in Chapter 16 about handson-ml HOT 3 CLOSED

def display_game_play(model_path, n_max_steps = 2000):
    env = gym.make('MsPacman-v0')
    obs = env.reset()
    with tf.Session() as sess:
        saver.restore(sess, model_path)
        for step in range(n_max_steps):
            #evaluate what to do
            state = preprocess_observation(obs)
            q_values = critic_q_values.eval(feed_dict={X_state: [state]})  #maybe actor_q_values?

            action_val = np.argmax(q_values)

            print("\rstep: {}, action_val: {}".format(step, action_val), end="")
            obs, reward, done, info = env.step(action_val)
            env.render()

            if done:
                print("DONE - FAILED")
                break
    env.close()
    return

save_path = "./data/ch16/my_dqn.ckpt"
display_game_play(save_path)