563acaf48cf475f4b125fd32374c6bb566049551,src/garage/torch/algos/vpg.py,VPG,process_samples,#VPG#Any#Any#,268

Before Change

        average_discounted_return = (np.mean(
            [path["returns"][0] for path in paths]))
        undiscounted_returns = [sum(path["rewards"]) for path in paths]
        average_return = np.mean(undiscounted_returns)

        self._episode_reward_mean.extend(undiscounted_returns)

        tabular.record("Iteration", itr)
        tabular.record("AverageDiscountedReturn", average_discounted_return)
        tabular.record("AverageReturn", average_return)
        tabular.record("Extras/EpisodeRewardMean",
                       np.mean(self._episode_reward_mean))
        tabular.record("NumTrajs", len(paths))
        tabular.record("StdReturn", np.std(undiscounted_returns))
        tabular.record("MaxReturn", np.max(undiscounted_returns))
        tabular.record("MinReturn", np.min(undiscounted_returns))

        return dict(average_return=average_return)

After Change

                                            axis=0) for path in paths
        ])
        rewards = torch.stack([
            loss_function_utils.pad_to_last(path["rewards"],
                                            total_length=self.max_path_length)
            for path in paths
        ])

        return valids, obs, actions, rewards

Instances