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

Safety Evaluation on SafetyBench

69.4Safety

Fine-tuned

34.35243.45152.5561.649Nov 11, 2025
Updated 2mo ago

Evaluation Results

MethodLinks
2025.11
69.4
2025.11
66.8
2025.11
66.5
2025.11
65.6
2025.11
65.2
2025.11
62.6
2025.11
61.9
2025.11
60.1
2025.11
57.5
2025.11
54.9
2025.11
54.8
2025.11
51.6
2025.11
50.3
2025.11
50
2025.11
49.7
2025.11
48.9
2025.11
48.2
2025.11
47.1
2025.11
46.5
2025.11
42.7
2025.11
37.5
2025.11
36.9
2025.11
36.6
2025.11
36.5
2025.11
35.9
2025.11
35.7