List your company

Safety Evaluation on XSTEST (HS, ASR)

2.05HS RateTRIDENT-EDGE
2.00882.28692.5652.8431May 30, 2025

Evaluation results

MethodLinks
TRIDENT-EDGEBackbone Model=META-LLAMA-3.1-8B2025.052.050.09
WILDBREAKBackbone Model=META-LLAMA-3.1-8B2025.052.10.07
AARTBackbone Model=META-LLAMA-3.1-8B2025.052.130.12
WILDCHATBackbone Model=META-LLAMA-3.1-8B2025.052.220.1
ATTAQBackbone Model=META-LLAMA-3.1-8B2025.052.310.15
SAFE_RLHFBackbone Model=META-LLAMA-3.1-8B2025.052.590.09
UnalignedBackbone Model=META-LLAMA-3.1-8B2025.052.770.32
HH_RLHFBackbone Model=META-LLAMA-3.1-8B2025.053.080.33