Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Direct Malicious on MM-SafetyBench OOD
Loading...
0.71
ASR
VLMShield
-1.3124
12.3388
25.99
39.6412
Apr 7, 2026
ASR
Updated 9d ago
Evaluation Results
Method
Method
Links
ASR
VLMShield
Base VLM=Qwen2.5-VL-7B...
2026.04
0.71
VLMShield
Base Model=LLaVA-1.5-13B
2026.04
0.71
ASTRA
Base VLM=Qwen2.5-VL-7B...
2026.04
4.12
ASTRA
Base Model=LLaVA-1.5-13B
2026.04
8.62
VLMGuard
Base VLM=Qwen2.5-VL-7B...
2026.04
9.05
VLMGuard
Base Model=LLaVA-1.5-13B
2026.04
12.9
ECSO
Base VLM=Qwen2.5-VL-7B...
2026.04
13.28
SelfReminder
Base VLM=Qwen2.5-VL-7B...
2026.04
15.2
ECSO
Base Model=LLaVA-1.5-13B
2026.04
18.39
JailGuard
Base VLM=Qwen2.5-VL-7B...
2026.04
24.35
MirrorCheck
Base VLM=Qwen2.5-VL-7B...
2026.04
25.08
MirrorCheck
Base Model=LLaVA-1.5-13B
2026.04
25.08
JailGuard
Base Model=LLaVA-1.5-13B
2026.04
36.22
CIDER
Base VLM=Qwen2.5-VL-7B...
2026.04
46.91
CIDER
Base Model=LLaVA-1.5-13B
2026.04
46.91
SelfReminder
Base Model=LLaVA-1.5-13B
2026.04
51.27
Feedback
Search any
task
Search any
task