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

Individual Label Prediction on HH Harmlessness consensus-eligible subset

79.8Per-individual Accuracy

Oracle consensus

43.71253.08162.4571.819Jul 7, 2026
Updated 18d ago

Evaluation Results

MethodLinks
2026.07
79.8
2026.07
50.2
2026.07
49
2026.07
45.1