Most of my ideas are not original but where can I learn more about this recursive reward modeling problem?