ResearchBenchmarksCommonsense Reasoning on aNLI (dev)Follow77.94AccuracyCKT75.44476.09276.7477.388Jun 4, 2023Evaluation ResultsMethodMethodLinksAccuracyCKTModel Scale=largeModel Scale=large2023.0677.94CALMModel Scale=largeModel Scale=large2023.0677.12T5Model Scale=largeModel Scale=large2023.0675.54