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

Harmful Content Detection on UnsafeBench

71.7AUPRC

Gemma

29.47640.43851.462.362May 1, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.05
71.7
2026.05
70.5
2026.05
69.7
2026.05
69.6
2026.05
69.1
2026.05
68.7
2026.05
68.7
2026.05
68.6
2026.05
67.6
2026.05
67.1
2026.05
67.1
2026.05
67
2026.05
66.8
2026.05
66.7
2026.05
66.5
2026.05
66.2
2026.05
66.1
2026.05
65.6
2026.05
65.5
2026.05
65.4
2026.05
64.2
2026.05
64.1
2026.05
64.1
2026.05
63.5
2026.05
63.4
2026.05
63.4
2026.05
63.2
2026.05
62.9
2026.05
61.9
2026.05
61.5
2026.05
61.3
2026.05
60.5
2026.05
57.8
2026.05
57.8
2026.05
57.4
2026.05
57.2
2026.05
55.6
2026.05
53
2026.05
52.2
2026.05
47.2
2026.05
40.4
2026.05
39.4
2026.05
37.4
2026.05
37.2
2026.05
35.6
2026.05
33.9
2026.05
33.9
2026.05
31.1
2026.05
31.1