ResearchBenchmarksQuestion Answering on TruthfulQA n=100Follow33MC1 ScoreHook ANC22.625.32830.7Mar 10, 2026Evaluation ResultsMethodMethodLinksMC1 ScoreMC2 ScoreToken F1 ScoreExact Match (EM)Hook ANCModel=LLaMA 3-8BModel=LLaMA 3-8B2026.033343.3521.830BaselineModel=Phi-3-miniModel=Phi-3-mini2026.032942.624.180BaselineModel=LLaMA 3-8BModel=LLaMA 3-8B2026.032943.0621.560Hook ANCModel=Phi-3-miniModel=Phi-3-mini2026.032842.523.70BaselineModel=OPT-125MModel=OPT-125M2026.032441.7917.810Hook ANCModel=OPT-125MModel=OPT-125M2026.032341.2617.280