Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Safety Evaluation on VLBreak (Challenge)
Loading...
0.04
ASR
VLGuard
-0.616
3.812
8.24
12.668
Aug 26, 2025
ASR
Updated 16d ago
Evaluation Results
Method
Method
Links
ASR
VLGuard
Backbone=Qwen2-VL
2025.08
0.04
PRISM
Backbone=Qwen2-VL
2025.08
0.05
PRISM
Backbone=LLaVA-1.5
2025.08
0.2
VLGuard
Backbone=LLaVA-1.5
2025.08
1.94
SPA-VL
Backbone=LLaVA-1.5
2025.08
6.54
SPA-VL
Backbone=Qwen2-VL
2025.08
7.82
SafeRLHF-V
Backbone=Qwen2-VL
2025.08
10.31
No Defense
Backbone=LLaVA-1.5
2025.08
13
No Defense
Backbone=Qwen2-VL
2025.08
15.12
SafeRLHF-V
Backbone=LLaVA-1.5
2025.08
16.44
Feedback
Search any
task
Search any
task