ICML 2026: "ProRL: Effective Reinforcement Learning for Proactive Recommendation via Rectified Policy Gradient Estimation"
hongruhou89/ProRL has added +20 stars since the first tracked point, with current momentum at 16.90.