Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Red-teaming on Violence prompts

0Failure Rate (FR)

UnlearnDiffAtk

-2.469214.197930.86547.5321Nov 25, 2024
Updated 21d ago

Evaluation Results

MethodLinks
2024.11
0
2024.11
0.74
2024.11
1.85
2024.11
3.2
2024.11
3.7
2024.11
4.05
2024.11
4.63
2024.11
4.8
2024.11
5.56
2024.11
5.6
2024.11
5.6
2024.11
5.6
2024.11
6.02
2024.11
6.02
2024.11
6.38
2024.11
6.94
2024.11
7.01
2024.11
7.41
2024.11
7.41
2024.11
9.72
2024.11
11.11
2024.11
11.57
2024.11
13.43
2024.11
13.89
2024.11
13.89
2024.11
14.35
2024.11
14.81
2024.11
15.74
2024.11
17.13
2024.11
17.14
2024.11
19.16
2024.11
22.69
2024.11
23.3
2024.11
24
2024.11
24.07
2024.11
27.78
2024.11
30.09
2024.11
30.56
2024.11
30.56
2024.11
36.57
2024.11
37.96
2024.11
38.89
2024.11
38.89
2024.11
42.75
2024.11
44.91
2024.11
47.45
2024.11
58.11
2024.11
61.73