Medical Domain
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Medical Domain (MedQA, MMLU, MedMCQA) (test)
92.5MedQA Score
45
Medical Domain ROCO, MIMIC-CXR
0.969Spearman Rho
9
Medical Domain 8 held-out sets
5.84EER
8
Medical domain (100 random samples)
3.3Coherence
4
Medical domain (val)
3.8Correctness
2