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

Harmfulness Detection on WildGuard

90.6Macro F1 Score

Qwen3Guard

6.110428.045249.9871.9148Apr 20, 2026Apr 26, 2026May 3, 2026May 9, 2026May 16, 2026May 22, 2026May 29, 2026
Updated 2d ago

Evaluation Results

MethodLinks
2026.04
90.6
2026.05
89.96
2026.05
89.8
2026.05
89.17
2026.04
89.1
2026.05
89.01
2026.05
88.9
2026.04
88.3
2026.04
86.7
2026.04
86.5
2026.04
86.5
2026.05
82.89
2026.04
81.3
2026.05
80.24
2026.05
80.13
2026.05
80.05
2026.05
79.82
2026.05
79.7
2026.05
78.71
2026.04
78.6
2026.05
78.5
2026.05
78.39
2026.05
78.2
2026.05
76.8
2026.05
76.37
2026.05
76.18
2026.05
75.93
2026.05
75.4
2026.05
74.48
2026.05
71.69
2026.05
71.5
2026.05
70.9
2026.05
70.8
2026.05
68.26
2026.05
67.44
2026.05
66.68
2026.05
66.5
2026.05
65.48
2026.05
57.74
2026.05
56.4
2026.05
49.1
2026.05
47
2026.05
32.36
2026.05
20.13
2026.05
16.9
2026.05
12.1
2026.05
9.36