dedc529d98ff3a6c82ffdf3baddb9a9edf569dc4,ch06/01_dqn_pong.py,,,#,207

Before Change

    writer = SummaryWriter(comment="-pong")
    env = BufferWrapper(ImageWrapper(gym.make("PongNoFrameskip-v4")), n_steps=4)

    test_env = BufferWrapper(ImageWrapper( gym.make("PongNoFrameskip-v4")), n_steps=4)
//    test_env = gym.wrappers.Monitor(test_env, "records", force=True)

    net = DQN(env.observation_space.shape, env.action_space.n)
    tgt_net = TargetNet(net, cuda=args.cuda)

After Change

        net.cuda()

    print("Populate buffer with %d steps" % REPLAY_START_SIZE)
    episode_rewards = []
    for _ in range(REPLAY_START_SIZE):
        rw = agent.play_step(None, epsilon=1.0)
        if rw is not None:
            episode_rewards.append(rw)

Instances