List your company

Jailbreak Attack Robustness on LogiBreak

0LogiBreak ASRSFT
-1.26.91523.1Sep 30, 2025

Evaluation results

MethodLinks
SFTratio=30/70, base_model=Llama-3.1-8B-Instruct2025.09059.31.5966.7015
RepBendbase_model=Llama-3.1-8B-Instruct2025.091368.778.968.208.5
ASGUARDbase_model=Llama-3.1-8B-Instruct2025.091397.164.968.174.745.8
Llama-3.1-8B-Instruct2025.093088.528.968.2——