"LLMs are (still) mostly powered by imitative learning, not RL" by Steven Byrnes

لـ LessWrong (Curated & Popular)