ResearchBenchmarksRecommendation on ML-20M (rating ≥ 4.5)Follow19.12NDCG@10Exp-RSFT1.27365.906810.5415.1732Mar 10, 2026Evaluation ResultsMethodMethodLinksNDCG@10NDCG@50HR@10HR@50MRRAverage RewardExp-RSFT2026.0319.1224.6232.3857.2616.513.8611RSFT2026.0318.4724.0331.6156.7715.933.7904BC2026.0317.5323.1330.3555.6915.13.7388PPO2026.0311.6416.8821.4145.3610.083.7547DPO2026.031.963.423.7210.521.954.4008