2b77a13f3870712284122f2a5411c7d4b8db82e6,yarll/agents/sac.py,SAC,learn,#SAC#,253

Before Change

                            self.train(model_summary=True)
                    state = new_state
                    if done:
                        summary = tf.Summary()
                        summary.value.add(tag="global/Episode_length",
                                          simple_value=float(episode_length))
                        summary.value.add(tag="global/Reward",
                                          simple_value=float(episode_reward))

After Change

                    state = new_state
                    if done:
                        self.total_episodes += 1
                        tf.summary.scalar("env/Episode_length", episode_length, self.total_steps)
                        tf.summary.scalar("env/Reward", episode_reward, self.total_steps)
                        tf.summary.scalar("env/N_episodes", self.total_episodes, step=self.total_steps)
                        break

Instances