MedSafetyBench
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
MedSafetyBench Hades
100Safety Score (1-ASR)
8
MedSafetyBench QR
100Safety Score
8
MedSafetyBench FigStep
0.998Safety Score (1-ASR)
8
MedSafetyBench Direct
100Safety Score
8
MedSafetyBench 100 sampled harmful queries
100ASR
2
MedSafetyBench
94.2Convergence Rate
2