ResearchBenchmarksSafety Alignment on PKU-SafeRLHF-10K (test)Follow356.32HV ScoreCAGE216.5856252.8628289.14325.4172May 8, 2026Evaluation ResultsMethodMethodLinksHV ScoreMIP ScoreCAGEBase Model=Alpaca-65B,...Base Model=Alpaca-65B, Reward Model Size=7B2026.05356.3282.8GenARMBase Model=Alpaca-65B,...Base Model=Alpaca-65B, Reward Model Size=7B2026.05262.2372.2PARMBase Model=Alpaca-65B,...Base Model=Alpaca-65B, Reward Model Size=7B2026.05221.9665.7