Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Multimodal Adversarial Attack on HADES
Loading...
91.33
Success Rate (Animal)
CS-DJ
81.6268
84.1459
86.665
89.1841
Jan 31, 2026
Success Rate (Animal)
Success Rate (Financial)
Success Rate (Privacy)
Success Rate (Self-Harm)
Success Rate (Violence)
Average Success Rate
Updated 4d ago
Evaluation Results
Method
Method
Links
Success Rate (Animal)
Success Rate (Financial)
Success Rate (Privacy)
Success Rate (Self-Harm)
Success Rate (Violence)
Average Success Rate
CS-DJ
Victim Model=Qwen3-VL-...
2026.01
91.33
88
81.33
56.67
94
82.27
Text-DJ
Victim Model=Qwen3-VL-...
2026.01
91.33
88
86
57.33
90
82.53
Text-DJ
Victim Model=Qwen3-VL-8B
2026.01
90
90
88.67
51.33
90.67
82.13
CS-DJ
Victim Model=Qwen3-VL-8B
2026.01
87.33
83.33
71.33
50.67
76.67
73.87
Text-DJ
Victim Model=Qwen3-VL-4B
2026.01
82.67
85.33
77.33
47.33
80.67
74.67
CS-DJ
Victim Model=Qwen3-VL-4B
2026.01
82
76.67
74.67
44.67
72.67
70.13
Feedback
Search any
task
Search any
task
