Accuracy percentage of different models. Our model achieved 82.92% accuracy on MedQA’s MCMLE, surpassing GPT-4’s 71.07%, highlighting the efficacy of our architecture and training methods [35, 36]. Our system achieved a 64.02% accuracy on the USMLE, lower than GPT-4’s 74.71%, primarily due to the evidence-based categorization in the datasets.
收藏NIAID Data Ecosystem2026-05-02 收录
下载链接：
https://figshare.com/articles/dataset/Accuracy_percentage_of_different_models_Our_model_achieved_82_92_accuracy_on_MedQA_s_MCMLE_surpassing_GPT-4_s_71_07_highlighting_the_efficacy_of_our_architecture_and_training_methods_35_36_Our_system_achieved_a_64_02_accuracy_on_the_USMLE_l/29934562
下载链接
链接失效反馈官方服务：
资源简介：
Accuracy percentage of different models. Our model achieved 82.92% accuracy on MedQA’s MCMLE, surpassing GPT-4’s 71.07%, highlighting the efficacy of our architecture and training methods [35, 36]. Our system achieved a 64.02% accuracy on the USMLE, lower than GPT-4’s 74.71%, primarily due to the evidence-based categorization in the datasets.
创建时间：
2025-08-18
