List your company

Helpfulness on PKU-SafeRLHF 30K

84.5Win RateSafeDPO
65.135270.162675.1980.2174May 26, 2025

Evaluation results

MethodLinks
SafeDPOBaseline=DPO-HELPFUL2025.0584.50.1315.38
SafeDPOBaseline=DPO-HARMLESS2025.0583.389.2516.38
SafeDPOBaseline=DPO-SAFEBETTER2025.0582.070.2517.68
SafeDPOBaseline=SACPO2025.05760.3823.63
SafeDPOBaseline=P-SACPO2025.0570.75029.25
SafeDPOBaseline=SafeRLHF2025.0565.88034.13