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

Response Harmfulness Detection on BeaverTails

89.9F1 Score

BeaverDam 7B

62.558469.656776.75583.8533Feb 3, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.02
89.9
2026.02
87.6
2026.02
86.7
2026.02
86.48
2026.02
86.04
2026.02
85.89
2026.02
85.35
2026.02
85.19
2026.02
84.99
2026.02
84.4
2026.02
79.39
2026.02
77.1
2026.02
75.2
2026.02
74.7
2026.02
73.8
2026.02
69.51
2026.02
67.84
2026.02
63.61