Qingyu Yin

Improving Sampling Efficiency in RLVR through Adaptive Rollout and Response Reuse

Longer Context, Deeper Thinking: Uncovering the Role of Long-Context Ability in Reasoning

Mitigating Heterogeneous Token Overfitting in LLM Knowledge Editing

Unlocking Efficient, Scalable, and Continual Knowledge Editing with Basis-Level Representation Fine-Tuning

Two-Step Offline Preference-Based Reinforcement Learning with Constrained Actions