Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

HarmB

Benchmarks

Task NameDataset NameSOTA ResultTrend
Harmful prompt detectionHarmB
F1 Score100
27
Harmful QueryHarmB
ASR (%)0.18
20
Prompt-side safety risk classificationHarmB
F1 Score100
12
Showing 3 of 3 rows