Hi there, for any one interested in healthcare and medicals, recently, we conducted a comprehensive accuracy test on Medical Chat, and the results exceeded our expectations. The accuracy rates for USMLE and MedQA reached 98.1% and 97.8%, respectively, surpassing any publicly available medical question-answering models, including Google's Med-PaLM 2 and Flan-PaLM. The testing methods and results are open-source, allowing anyone to replicate them. If you're interested, feel free to check out this evaluation report. This is the open source repo for the testing , anyone can reproduce the testing results.