DQN LunarLander.py

class LunarLanderDQN(DQN):

def __init__(
            self,
            env: str,
            eps_last_frame: int = 10000,
            sync_rate: int = 10,
            learning_rate: float = 1e-2,
            batch_size: int = 16,
            replay_size: int = 10000,
            warm_start_size: int = 10000,
            **kwargs,
    ):
    super().__init__(env, eps_last_frame, sync_rate, learning_rate,
                     replay_size, warm_start_size, **kwargs)
    
    # New environment
    self.env = gym.make(env)
    
    self.obs_shape = self.env.observation_space.shape
    self.n_actions = self.env.action_space.n
    
    # New network
    self.net = MLP(self.obs_shape, self.n_actions)
    self.target_net = MLP(self.obs_shape, self.n_actions)