for episode in range(num_episodes):
        state, _ = self.env.reset()
        done = False
        total_reward = 0
        log_probs = []
        rewards = []