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

Jailbreak Attack on Prefilling Attack 10 tokens

70.91ASR

SD_DaExpert

-2.836416.309335.45554.6007Feb 2, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.02
70.914.54
2026.02
66.674.48
2026.02
65.764.38
2026.02
53.944.17
2026.02
52.424.05
2026.02
504.06
2026.02
503.95
2026.02
29.73.22
2026.02
29.392.65
2026.02
26.672.61
2026.02
26.673.05
2026.02
26.362.87
2026.02
26.362.95
2026.02
21.822.88
2026.02
19.392.42
2026.02
16.972.38
2026.02
15.762.19
2026.02
15.152.28
2026.02
14.851.89
2026.02
14.852.33
2026.02
13.942.08
2026.02
13.032.31
2026.02
11.821.47
2026.02
9.391.78
2026.02
9.091.44
2026.02
8.791.7
2026.02
8.481.62
2026.02
8.181.78
2026.02
7.271.33
2026.02
6.061.45
2026.02
4.851.53
2026.02
4.851.89
2026.02
4.551.77
2026.02
4.241.47
2026.02
4.241.27
2026.02
3.331.28
2026.02
3.031.42
2026.02
2.731.16
2026.02
2.731.16
2026.02
0.911.06
2026.02
0.911.09
2026.02
0.611.04
2026.02
0.611.15
2026.02
0.31.34
2026.02
01.05