Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Multimodal Jailbreaking on SafeBench Mirror (ID)
Loading...
100
ASR
Vanilla
-4
23
50
77
Mar 2, 2026
ASR
Mean Judge Score
Updated 1mo ago
Evaluation Results
Method
Method
Links
ASR
Mean Judge Score
Vanilla
Model=LLaVA-1.5-7B
2026.03
100
4.33
Vanilla
Model=Qwen3-VL-8B
2026.03
90.9
4.55
Static SFT
Model=Qwen3-VL-8B
2026.03
0
0
CEMMA
Model=Qwen3-VL-8B
2026.03
0
0
Static SFT
Model=LLaVA-1.5-7B
2026.03
0
0
CEMMA
Model=LLaVA-1.5-7B
2026.03
0
0
Feedback
Search any
task
Search any
task