Mark end of episode
(self)
| 666 | self.curr_ep_loss_length += 1 |
| 667 | |
| 668 | def log_episode(self): |
| 669 | "Mark end of episode" |
| 670 | self.ep_rewards.append(self.curr_ep_reward) |
| 671 | self.ep_lengths.append(self.curr_ep_length) |
| 672 | if self.curr_ep_loss_length == 0: |
| 673 | ep_avg_loss = 0 |
| 674 | ep_avg_q = 0 |
| 675 | else: |
| 676 | ep_avg_loss = np.round(self.curr_ep_loss / self.curr_ep_loss_length, 5) |
| 677 | ep_avg_q = np.round(self.curr_ep_q / self.curr_ep_loss_length, 5) |
| 678 | self.ep_avg_losses.append(ep_avg_loss) |
| 679 | self.ep_avg_qs.append(ep_avg_q) |
| 680 | |
| 681 | self.init_episode() |
| 682 | |
| 683 | def init_episode(self): |
| 684 | self.curr_ep_reward = 0.0 |
no test coverage detected