ResearchBenchmarksReward Modeling on UltraFeedback CleanedFollow92.36Total ScoreVRM53.235263.392673.5583.7074Mar 5, 2026Evaluation ResultsMethodMethodLinksTotal ScoreVRM2026.0392.36RM2026.0388.98SELECTIVE DPO2026.0382.28DPO2026.0375.71KTO2026.0370.67IPO2026.0360.69SIMPO2026.0359.16WPO2026.0354.74