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

Unsafe Robustness on AdvBench

0Unsafe Rate

LLaDA-Instruct

-3.37619.41242.264.988Apr 28, 2026
Updated 22d ago

Evaluation Results

MethodLinks
2026.04
099.6
2026.04
0-
2026.04
0-
2026.04
0-
2026.04
099.4
2026.04
033.9
2026.04
0-
2026.04
0-
2026.04
0-
2026.04
0-
2026.04
0-
2026.04
0-
2026.04
0-
2026.04
0-
2026.04
099.6
2026.04
0-
2026.04
0-
2026.04
0-
2026.04
099.2
2026.04
039.6
2026.04
0-
2026.04
0-
2026.04
0-
2026.04
0-
2026.04
0-
2026.04
0-
2026.04
090.6
2026.04
047.1
2026.04
0-
2026.04
0-
2026.04
0-
2026.04
0.2-
2026.04
0.4-
2026.04
0.4-
2026.04
43.5
2026.04
4.2-
2026.04
4.4-
2026.04
4.64
2026.04
4.6-
2026.04
4.6-
2026.04
5-
2026.04
6-
2026.04
13.5-
2026.04
13.7-
2026.04
14-
2026.04
14.8-
2026.04
17.7-
2026.04
20-
2026.04
24.645.4
2026.04
28.1-
2026.04
31.5-
2026.04
31.7-
2026.04
32.557.9
2026.04
32.759
2026.04
34-
2026.04
35.4-
2026.04
35.8-
2026.04
36.3-
2026.04
39.4-
2026.04
4437.3
2026.04
44.8-
2026.04
46.9-
2026.04
61.9-
2026.04
62.10.2
2026.04
63.8-
2026.04
71.2-
2026.04
71.90
2026.04
72.7-
2026.04
76.20.2
2026.04
76.7-
2026.04
82.30.2
2026.04
84.4-