Individual Label Prediction on HH Helpfulness (consensus-eligible subset)

76.7Per-individual Accuracy

Oracle consensus

Updated 18d ago

Evaluation Results

Method	Links
Oracle consensus 2026.07		76.7
PERSONAJUDGE 2026.07		52.7
Per-evaluator majority-class 2026.07		46.4
Global majority-class 2026.07		34.3