GPG: A Simple and Strong Reinforcement Learning Baseline for Model Reasoning Paper • 2504.02546 • Published Apr 3, 2025 • 2