Well they can be used together in some contexts so while they are different, you could also say RL can help Supervised Fine Tuning for further optimization