How to use the tianshou.env.SubprocVectorEnv function in tianshou

def test_a2c(args=get_args()):
    env = gym.make(args.task)
    args.state_shape = env.observation_space.shape or env.observation_space.n
    args.action_shape = env.action_space.shape or env.action_space.n
    # train_envs = gym.make(args.task)
    train_envs = SubprocVectorEnv(
        [lambda: gym.make(args.task) for _ in range(args.training_num)],
        reset_after_done=True)
    # test_envs = gym.make(args.task)
    test_envs = SubprocVectorEnv(
        [lambda: gym.make(args.task) for _ in range(args.test_num)],
        reset_after_done=False)
    # seed
    np.random.seed(args.seed)
    torch.manual_seed(args.seed)
    train_envs.seed(args.seed)
    test_envs.seed(args.seed)
    # model
    net = Net(args.layer_num, args.state_shape, device=args.device)
    actor = Actor(net, args.action_shape).to(args.device)
    critic = Critic(net).to(args.device)
    optim = torch.optim.Adam(list(

def test_a2c(args=get_args()):
    env = gym.make(args.task)
    args.state_shape = env.observation_space.shape or env.observation_space.n
    args.action_shape = env.action_space.shape or env.action_space.n
    # train_envs = gym.make(args.task)
    train_envs = SubprocVectorEnv(
        [lambda: gym.make(args.task) for _ in range(args.training_num)],
        reset_after_done=True)
    # test_envs = gym.make(args.task)
    test_envs = SubprocVectorEnv(
        [lambda: gym.make(args.task) for _ in range(args.test_num)],
        reset_after_done=False)
    # seed
    np.random.seed(args.seed)
    torch.manual_seed(args.seed)
    train_envs.seed(args.seed)
    test_envs.seed(args.seed)
    # model
    net = Net(args.layer_num, args.state_shape, device=args.device)
    actor = Actor(net, args.action_shape).to(args.device)
    critic = Critic(net).to(args.device)
    optim = torch.optim.Adam(list(
        actor.parameters()) + list(critic.parameters()), lr=args.lr)
    dist = torch.distributions.Categorical
    policy = A2CPolicy(
        actor, critic, optim, dist, args.gamma, vf_coef=args.vf_coef,

def test_vecenv(size=10, num=8, sleep=0.001):
    verbose = __name__ == '__main__'
    env_fns = [
        lambda i=i: MyTestEnv(size=i, sleep=sleep)
        for i in range(size, size + num)
    ]
    venv = [
        VectorEnv(env_fns),
        SubprocVectorEnv(env_fns),
    ]
    if verbose:
        venv.append(RayVectorEnv(env_fns))
    for v in venv:
        v.seed()
    action_list = [1] * 5 + [0] * 10 + [1] * 20
    if not verbose:
        o = [v.reset() for v in venv]
        for i, a in enumerate(action_list):
            o = []
            for v in venv:
                A, B, C, D = v.step([a] * num)
                if sum(C):
                    A = v.reset(np.where(C)[0])
                o.append([A, B, C, D])
            for i in zip(*o):

def test_sac(args=get_args()):
    torch.set_num_threads(1)
    env = gym.make(args.task)
    args.state_shape = env.observation_space.shape or env.observation_space.n
    args.action_shape = env.action_space.shape or env.action_space.n
    args.max_action = env.action_space.high[0]
    # you can also use tianshou.env.SubprocVectorEnv
    # train_envs = gym.make(args.task)
    train_envs = SubprocVectorEnv(
        [lambda: gym.make(args.task) for _ in range(args.training_num)])
    # test_envs = gym.make(args.task)
    test_envs = SubprocVectorEnv(
        [lambda: gym.make(args.task) for _ in range(args.test_num)])
    # seed
    np.random.seed(args.seed)
    torch.manual_seed(args.seed)
    train_envs.seed(args.seed)
    test_envs.seed(args.seed)
    # model
    net = Net(args.layer_num, args.state_shape, device=args.device)
    actor = ActorProb(
        net, args.action_shape,
        args.max_action, args.device, unbounded=True
    ).to(args.device)
    actor_optim = torch.optim.Adam(actor.parameters(), lr=args.actor_lr)

def test_a2c(args=get_args()):
    env = create_atari_environment(args.task)
    args.state_shape = env.observation_space.shape or env.observation_space.n
    args.action_shape = env.env.action_space.shape or env.env.action_space.n
    # train_envs = gym.make(args.task)
    train_envs = SubprocVectorEnv(
        [lambda: create_atari_environment(args.task)
         for _ in range(args.training_num)])
    # test_envs = gym.make(args.task)
    test_envs = SubprocVectorEnv(
        [lambda: create_atari_environment(args.task)
         for _ in range(args.test_num)])
    # seed
    np.random.seed(args.seed)
    torch.manual_seed(args.seed)
    train_envs.seed(args.seed)
    test_envs.seed(args.seed)
    # model
    net = Net(args.layer_num, args.state_shape, device=args.device)
    actor = Actor(net, args.action_shape).to(args.device)
    critic = Critic(net).to(args.device)
    optim = torch.optim.Adam(list(
        actor.parameters()) + list(critic.parameters()), lr=args.lr)
    dist = torch.distributions.Categorical
    policy = A2CPolicy(
        actor, critic, optim, dist, args.gamma, vf_coef=args.vf_coef,

def test_ddpg(args=get_args()):
    env = gym.make(args.task)
    args.state_shape = env.observation_space.shape or env.observation_space.n
    args.action_shape = env.action_space.shape or env.action_space.n
    args.max_action = env.action_space.high[0]
    # train_envs = gym.make(args.task)
    train_envs = VectorEnv(
        [lambda: gym.make(args.task) for _ in range(args.training_num)])
    # test_envs = gym.make(args.task)
    test_envs = SubprocVectorEnv(
        [lambda: gym.make(args.task) for _ in range(args.test_num)])
    # seed
    np.random.seed(args.seed)
    torch.manual_seed(args.seed)
    train_envs.seed(args.seed)
    test_envs.seed(args.seed)
    # model
    net = Net(args.layer_num, args.state_shape, device=args.device)
    actor = Actor(net, args.action_shape, args.max_action,
                  args.device).to(args.device)
    actor_optim = torch.optim.Adam(actor.parameters(), lr=args.actor_lr)
    net = Net(args.layer_num, args.state_shape,
              args.action_shape, concat=True, device=args.device)
    critic = Critic(net, args.device).to(args.device)
    critic_optim = torch.optim.Adam(critic.parameters(), lr=args.critic_lr)
    policy = DDPGPolicy(

def test_sac(args=get_args()):
    env = gym.make(args.task)
    args.state_shape = env.observation_space.shape or env.observation_space.n
    args.action_shape = env.action_space.shape or env.action_space.n
    args.max_action = env.action_space.high[0]
    # train_envs = gym.make(args.task)
    train_envs = VectorEnv(
        [lambda: gym.make(args.task) for _ in range(args.training_num)])
    # test_envs = gym.make(args.task)
    test_envs = SubprocVectorEnv(
        [lambda: gym.make(args.task) for _ in range(args.test_num)])
    # seed
    np.random.seed(args.seed)
    torch.manual_seed(args.seed)
    train_envs.seed(args.seed)
    test_envs.seed(args.seed)
    # model
    net = Net(args.layer_num, args.state_shape, device=args.device)
    actor = ActorProb(
        net, args.action_shape,
        args.max_action, args.device, unbounded=True
    ).to(args.device)
    actor_optim = torch.optim.Adam(actor.parameters(), lr=args.actor_lr)
    net = Net(args.layer_num, args.state_shape,
              args.action_shape, concat=True, device=args.device)
    critic1 = Critic(net, args.device).to(args.device)

def test_sac(args=get_args()):
    torch.set_num_threads(1)
    env = gym.make(args.task)
    args.state_shape = env.observation_space.shape or env.observation_space.n
    args.action_shape = env.action_space.shape or env.action_space.n
    args.max_action = env.action_space.high[0]
    # you can also use tianshou.env.SubprocVectorEnv
    # train_envs = gym.make(args.task)
    train_envs = SubprocVectorEnv(
        [lambda: gym.make(args.task) for _ in range(args.training_num)])
    # test_envs = gym.make(args.task)
    test_envs = SubprocVectorEnv(
        [lambda: gym.make(args.task) for _ in range(args.test_num)])
    # seed
    np.random.seed(args.seed)
    torch.manual_seed(args.seed)
    train_envs.seed(args.seed)
    test_envs.seed(args.seed)
    # model
    net = Net(args.layer_num, args.state_shape, device=args.device)
    actor = ActorProb(
        net, args.action_shape,
        args.max_action, args.device, unbounded=True
    ).to(args.device)
    actor_optim = torch.optim.Adam(actor.parameters(), lr=args.actor_lr)
    net = Net(args.layer_num, args.state_shape,
              args.action_shape, concat=True, device=args.device)
    critic1 = Critic(net, args.device).to(args.device)

def test_a2c(args=get_args()):
    env = create_atari_environment(args.task)
    args.state_shape = env.observation_space.shape or env.observation_space.n
    args.action_shape = env.env.action_space.shape or env.env.action_space.n
    # train_envs = gym.make(args.task)
    train_envs = SubprocVectorEnv(
        [lambda: create_atari_environment(args.task)
         for _ in range(args.training_num)])
    # test_envs = gym.make(args.task)
    test_envs = SubprocVectorEnv(
        [lambda: create_atari_environment(args.task)
         for _ in range(args.test_num)])
    # seed
    np.random.seed(args.seed)
    torch.manual_seed(args.seed)
    train_envs.seed(args.seed)
    test_envs.seed(args.seed)
    # model
    net = Net(args.layer_num, args.state_shape, device=args.device)
    actor = Actor(net, args.action_shape).to(args.device)
    critic = Critic(net).to(args.device)
    optim = torch.optim.Adam(list(

def test_td3(args=get_args()):
    reg()
    env = gym.make(args.task)
    args.state_shape = env.observation_space.shape or env.observation_space.n
    args.action_shape = env.action_space.shape or env.action_space.n
    args.max_action = env.action_space.high[0]
    # train_envs = gym.make(args.task)
    train_envs = VectorEnv(
        [lambda: gym.make(args.task) for _ in range(args.training_num)])
    # test_envs = gym.make(args.task)
    test_envs = SubprocVectorEnv(
        [lambda: gym.make(args.task) for _ in range(args.test_num)])
    # seed
    np.random.seed(args.seed)
    torch.manual_seed(args.seed)
    train_envs.seed(args.seed)
    test_envs.seed(args.seed)
    # model
    net = Net(args.layer_num, args.state_shape, device=args.device)
    actor = Actor(
        net, args.action_shape,
        args.max_action, args.device
    ).to(args.device)
    actor_optim = torch.optim.Adam(actor.parameters(), lr=args.actor_lr)
    net = Net(args.layer_num, args.state_shape,
              args.action_shape, concat=True, device=args.device)
    critic1 = Critic(net, args.device).to(args.device)