List your company

Alignment on Human Values (Help, Honest, Truth)

0.7165Helpfulness ScoreMAH-DPO
0.5998120.6301060.66040.690694Oct 1, 2025

Evaluation results

MethodLinks
MAH-DPOPRM Guidance=Help PRM2025.100.71650.45540.389
MAH-DPOPRM Guidance=Honest PRM2025.100.69680.51960.4107
MAH-DPOPRM Guidance=Truth PRM2025.100.68340.48720.363
MAH-DPO2025.100.63890.36870.2478
MODPO2025.100.61750.34770.2325
Single-Head DPO2025.100.60430.30550.2014