2 ms·This sounds like reward modeling combined with RLHF.by kmeisthax 3y agoThis sounds like reward modeling combined with RLHF.