Apple and Duke Researchers Present a Reinforcement Learning Approach That Enables LLMs to Provide Intermediate Answers, Enhancing Speed and Accuracy 发表评论 / 作者: / 5 月 30, 2025