def __init__(self, device: TorchDevice) -> None:
env_raw = make_atari("BreakoutNoFrameskip-v4")
self.__env_train = wrap_deepmind(env_raw, episode_life=True)
env_raw = make_atari("BreakoutNoFrameskip-v4")
self.__env_eval = wrap_deepmind(env_raw, episode_life=True)
self.__env = self.__env_train
self.__device = device
def __init__(self, device: TorchDevice) -> None:
env_raw = make_atari("BreakoutNoFrameskip-v4") # 加载预有的breakout模型
self.__env_train = wrap_deepmind(env_raw, episode_life=True) # 训练环境
env_raw = make_atari("BreakoutNoFrameskip-v4") # 加载预有的breakout模型
self.__env_eval = wrap_deepmind(env_raw, episode_life=True) # 测试环境
self.__env = self.__env_train # 默认为训练环境
self.__device = device