Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
MLLM Jailbreaking on MM-SafetyBench Physical Harm scenario
Loading...
6
ASR
Vanilla
2.48
26.24
50
73.76
Mar 26, 2025
ASR
Updated 4d ago
Evaluation Results
Method
Method
Links
ASR
Vanilla
Target Model=o1
2025.03
6
FigStep
Target Model=o1
2025.03
6
FigStep-Pro
Target Model=o1
2025.03
6
HADES
Target Model=o1
2025.03
6
FigStep-Pro
Target Model=GPT-4o
2025.03
10
Vanilla
Target Model=GPT-4o
2025.03
16
FigStep
Target Model=GPT-4o
2025.03
16
HADES
Target Model=GPT-4o
2025.03
19
FigStep
Target Model=Qwen2-VL-7B
2025.03
52
JOOD
Target Model=o1
2025.03
52
Vanilla
Target Model=Qwen2-VL-7B
2025.03
55
FigStep-Pro
Target Model=Qwen2-VL-7B
2025.03
68
JOOD
Target Model=GPT-4o
2025.03
74
HADES
Target Model=Qwen2-VL-7B
2025.03
79
JOOD
Target Model=Qwen2-VL-7B
2025.03
94
Feedback
Search any
task
Search any
task