ResearchBenchmarksMassive Multitask Language Understanding on MMLU (en)Follow23.5AccuracyBPE22.9823.11523.2523.385Jul 17, 2025Evaluation ResultsMethodMethodLinksAccuracyBPEmode=Zero-shotmode=Zero-shot2025.0723.5FXT λ3mode=Zero-shotmode=Zero-shot2025.0723.3BINOMIALmode=Zero-shotmode=Zero-shot2025.0723