ResearchBenchmarksLanguage Modeling on MMLU (test)Follow7.4797Anatomy AccuracyRegular5.8121646.2450826.6787.110918Jun 29, 2026Evaluation ResultsMethodMethodLinksAnatomy AccuracyAstronomy AccuracyBusiness Ethics AccuracyClinical Knowledge AccuracyAbstract Algebra AccuracyAverage Accuracy (MMLU)RegularBackbone=LLAMA-2-7BBackbone=LLAMA-2-7B2026.067.47977.77488.05117.56118.56037.8854GREATSBackbone=LLAMA-2-7BBackbone=LLAMA-2-7B2026.067.22786.17856.50146.14077.4426.6981GAIABackbone=LLAMA-2-7BBackbone=LLAMA-2-7B2026.065.87636.28965.53515.27916.62055.9201