ResearchBenchmarksNatural Language Inference on AutoTNLIFollow72.5AccuracyCR49.155.17561.2567.325Aug 8, 2023Evaluation ResultsMethodMethodLinksAccuracyCRModel=LLaMA-65B, n=4Model=LLaMA-65B, n=42023.0872.5CoTModel=LLaMA-65BModel=LLaMA-65B2023.0863.2CoT-SCModel=LLaMA-65B, k=16Model=LLaMA-65B, k=162023.0861.7DirectModel=LLaMA-65BModel=LLaMA-65B2023.0859.7CRModel=LLaMA-13B, n=4Model=LLaMA-13B, n=42023.0857CoTModel=LLaMA-13BModel=LLaMA-13B2023.0854.1DirectModel=LLaMA-13BModel=LLaMA-13B2023.0852.6CoT-SCModel=LLaMA-13B, k=16Model=LLaMA-13B, k=162023.0852.1RandomModel=-Model=-2023.0850