List your company

False Alarm Detection on AlpacaEval

0False Alarm Rate (FAR)Llama-Guard-3
-1.528.741929.26May 11, 2026

Evaluation results

MethodLinks
Llama-Guard-3Target Model=Vicuna2026.050
Llama-Guard-3Target Model=Llama22026.050
Llama-Guard-3Target Model=Qwen2.52026.050
embedding disruption methodTarget Model=Qwen2.52026.050
embedding disruption methodTarget Model=Vicuna2026.051
embedding disruption methodTarget Model=Llama22026.051
GuardReasonerTarget Model=Vicuna2026.0538
GuardReasonerTarget Model=Llama22026.0538
GuardReasonerTarget Model=Qwen2.52026.0538