List your company

Alignment-diversity coverage on HH-RLHF

81.7CoverageEvoPref
32.8245.5158.270.89May 21, 2026

Evaluation results

MethodLinks
EvoPref2026.0581.70.830.18
Group DPO2026.0568.40.840.24
MO-RLHF2026.0563.80.850.29
DPO-Ensembleensemble size=282026.0554.30.860.41
DPOtype=single2026.0534.70.880.65