finish a trajectory :param done: whether the epoch is done :param next_state: next state :return: None
(self, done, next_state)
| 423 | self.train_vf(states, rewards_to_go) |
| 424 | |
| 425 | def finish_path(self, done, next_state): |
| 426 | """ |
| 427 | finish a trajectory |
| 428 | :param done: whether the epoch is done |
| 429 | :param next_state: next state |
| 430 | :return: None |
| 431 | """ |
| 432 | if not done: |
| 433 | next_state = np.array([next_state], np.float32) |
| 434 | last_val = self.critic(next_state) |
| 435 | else: |
| 436 | last_val = 0 |
| 437 | self.buf.finish_path(last_val) |
| 438 | |
| 439 | |
| 440 | if __name__ == '__main__': |