render = False if episode % SHOW_INFOS == 1: show_infos(episode) init_game() over = False if render: show_render() while not over: action = policy.choose_action() env.apply_action(action) over = is_over() policy.update_replay_memory(over) policy.train(over) steps_remaining -= 1 if render: show_render() if policy.END_EPSILON_DECAYING >= episode >= policy.START_EPSILON_DECAYING: policy.epsilon -= policy.epsilon_decay_value policy.test_model() draw_test(windowSurface)