[논문리뷰] STAPO: Stabilizing Reinforcement Learning for LLMs by Silencing Rare Spurious TokensZhilong Zheng이 arXiv에 게시한 'STAPO: Stabilizing Reinforcement Learning for LLMs by Silencing Rare Spurious Tokens' 논문에 대한 자세한 리뷰입니다.#Review#Reinforcement Learning#Large Language Models#Training Stability#Policy Optimization#Spurious Tokens#Entropy Regularization#Gradient Modulation2026년 2월 17일댓글 수 로딩 중