medical QA
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Medical QA Multi-choice
70.7MMLU-Med Accuracy
22
Out-of-domain medical QA History, Engineering, Law (test)
50.4History Accuracy
10
Medical QA
92.5GPT-4 Score
9
Medical QA Primary n=319 (test)
21.94UCCR
8
Medical QA Primary Split n = 314 (dev)
20.7UCCR
8
Medical QA offline evaluation
0.83Honesty
3