c829d1a0bc7ded09ed0057ad4bc8493b3ab12fa0,ch16/02_breakout_es.py,,,#,176

Before Change

        dt_data = time.time() - t_start
        m_reward = np.mean(batch_reward)
        if m_reward > 18:
            print("Solved in %d steps" % step_idx)
            break

        train_step(optimizer, net, batch_noise, batch_reward, writer, step_idx)
        writer.add_scalar("reward_mean", m_reward, step_idx)
        writer.add_scalar("reward_std", np.std(batch_reward), step_idx)
        writer.add_scalar("reward_max", np.max(batch_reward), step_idx)

After Change

            step_idx, m_reward, speed, dt_data, dt_step, np.mean(batch_steps_data),
            np.min(batch_steps_data), np.max(batch_steps_data), np.std(batch_steps_data)))

    for worker, p_queue in zip(workers, params_queues):
        p_queue.put(None)
        worker.join()

Instances