ResearchBenchmarksReward Modeling on RewardBench v2 (Score and Rank)Follow57ScoreRBS32.24838.67445.151.526Feb 11, 2026Evaluation ResultsMethodMethodLinksScoreRankRBSBackbone=Qwen2.5-7B-Inst.Backbone=Qwen2.5-7B-Inst.2026.02575RBSBackbone=Qwen2.5-3B-Inst.Backbone=Qwen2.5-3B-Inst.2026.0246.28RBSBackbone=Llama-3.1-3BBackbone=Llama-3.1-3B2026.0236.29RBSBackbone=Llama-3.1-1BBackbone=Llama-3.1-1B2026.0233.23