Announcement_2

Paper accepted at ICLR 2026: “On the Generalization of SFT: A Reinforcement Learning Perspective with Reward Rectification.”