Accelerating RL for LLM Reasoning with Optimal Advantage Regression Paper • 2505.20686 • Published May 27 • 2