Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Safety Alignment on VSA Violence (visual synonym attack prompts)
Loading...
0
ASR
AEGIS
-2.6
14.95
32.5
50.05
Jul 7, 2026
ASR
Updated 17d ago
Evaluation Results
Method
Method
Links
ASR
AEGIS
Category=AEGIS (Ours),...
2026.07
0
AEGIS
Category=AEGIS (Ours),...
2026.07
4
SafeClip
Category=Input-Space
2026.07
5
SLD-Strong
Category=Trigger-Specific
2026.07
5
SAFREE
Category=Input-Space
2026.07
8
SafeGuider
Category=Input-Space
2026.07
11
CAD
Category=Feature Pruning
2026.07
12
UCE
Category=Trigger-Specific
2026.07
15
Receler
Category=Trigger-Specific
2026.07
17
SDID
Category=Trigger-Specific
2026.07
21
ESD-u
Category=Trigger-Specific
2026.07
30
SLD-Medium
Category=Trigger-Specific
2026.07
32
Diff-QuickFix
Category=Trigger-Specific
2026.07
45
SD 2.1
Category=No Alignment
2026.07
51
SLD-Weak
Category=Trigger-Specific
2026.07
54
SD 1.4
Category=No Alignment
2026.07
65
Feedback
Search any
task
Search any
task