List your company

Jailbreaking Success Rate on StrongREJECT (Detoxify & Llama Guard 3)

0ASR (Detoxify)VAE-JB
-0.5243.0136.5510.087Apr 11, 2026

Evaluation results

MethodLinks
VAE-JBModel=InternVL, epsilon (ε)=322026.0403.8
BAPModel=LLaVA-1.5, epsilon (ε)=322026.04047
BAPModel=InternVL, epsilon (ε)=322026.0403.5
JB-PcsModel=InternVL, epsilon (ε)=–2026.0404.5
JB-PcsModel=Qwen-VL, epsilon (ε)=–2026.040.35.8
Attention-Guided Visual JailbreakingModel=MiniGPT-4, epsilon (ε)=32, transfer=true2026.040.366.1
BAPModel=Qwen-VL, epsilon (ε)=322026.040.69.3
BAPModel=MiniGPT-4, epsilon (ε)=32, transfer=true2026.040.664.3
JB-PcsModel=LLaVA-1.5, epsilon (ε)=–2026.040.650.2
Attention-Guided Visual JailbreakingModel=MiniGPT-4, epsilon (ε)=16, transfer=true2026.040.664.9
VAE-JBModel=LLaVA-1.5, epsilon (ε)=322026.04151.8
VAE-JBModel=MiniGPT-4, epsilon (ε)=32, transfer=true2026.04164.5
JB-PcsModel=MiniGPT-4, epsilon (ε)=–, transfer=true2026.04163.9
Attention-Guided Visual JailbreakingModel=LLaVA-1.5, epsilon (ε)=162026.042.659.7
VAE-JBModel=Qwen-VL, epsilon (ε)=322026.042.955.6
Attention-Guided Visual JailbreakingModel=Qwen-VL, epsilon (ε)=162026.042.966.5
Attention-Guided Visual JailbreakingModel=InternVL, epsilon (ε)=162026.045.115.3
Attention-Guided Visual JailbreakingModel=InternVL, epsilon (ε)=322026.045.87
Attention-Guided Visual JailbreakingModel=LLaVA-1.5, epsilon (ε)=322026.048.578
Attention-Guided Visual JailbreakingModel=Qwen-VL, epsilon (ε)=322026.0413.190.4