HADES
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
HADES
99.6Attack Success Rate
132
HADES
72.66ASR
32
HADES
100Refusal Rate
25
HADES
3.6ASR
24
HADES
66.67Success Rate (Animal)
23
HADES (test)
64.815Ensemble Score
22
HADES-Dataset
99.07ASR (%)
20
HADES SD+TYPO
2.1Attack Success Rate (ASR)
18
HADES SD
2.8ASR
18
HADES OOD
80.02Attack Success Rate (ASR)
16
HADES All categories
18.93ASR
15
HADES Violence
0.3ASR
15
HADES Animals
3.33ASR
15
HADES Financial
95.33ASR
15
HADES Privacy
95.33ASR
15
HADES Self-harm
5.33ASR
15
HADES (test)
96Self-harm Success Rate
15
HADES
87Jailbreak Success Rate (Qwen2.5-VL-7B)
12
HADES
76StrongREJECT
10
HADES
91.33Success Rate (Animal)
6
HADES-Dataset
43.67HADES Score
4