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

Response-side Safety Benchmarks

Benchmarks

Task NameDataset NameSOTA ResultTrend
Safety GuardrailResponse-side safety benchmarks
F1 Score87
12
Response-side Safety ClassificationResponse-side Safety Benchmarks HarmB, SafeRLHF, BeaverTails, XSTest, Aegis2, WildG, Think
HarmB Score98.5
12
Showing 2 of 2 rows