AI RESEARCH
RewardBench 2: Advancing Reward Model Evaluation
arXiv CS.CL
•
ArXi:2506.01937v2 Announce Type: replace Reward models are used throughout the post-