List your company

Helpful Assistant Evaluation (Helpfulness and Harmlessness) on HH-RLHF

1.98Helpfulness Value (HV)MI-EPO
0.39920.80961.221.6304Jul 1, 2026

Evaluation results

MethodLinks
MI-EPOBackbone=Qwen3-8B-Base2026.071.98680.52
MO-ODPOBackbone=Qwen3-8B-Base2026.071.68640.57
RSBackbone=Qwen3-8B-Base2026.070.68210.58
RiCBackbone=Qwen3-8B-Base2026.070.4671.05