Zhaoyi Zhou

Variance-Reduced Reinforcement Learning for Large Reasoning Models via James-Stein Baselines

Accelerating Unbiased LLM Evaluation via Synthetic Feedback