Chujie Zheng

BARREL: Boundary-Aware Reasoning for Factual and Reliable LRMs

WorldPM: Understanding Scaling Patterns in Human Preference Modeling

From Theft to Bomb-Making: The Ripple Effect of Unlearning in Defending Against Jailbreak Attacks

Beyond the 80/20 Rule: High-Entropy Minority Tokens Drive Effective Reinforcement Learning for LLM Reasoning

Model Extrapolation Expedites Alignment