List your company

Combined win rate evaluation on PKU-SafeRLHF prompts n = 100 samples (Sev-Low)

53.6CVaR(0.125) Combined Win RateIPO Entropic τ=10
0.35214.1762841.824May 11, 2026

Evaluation results

MethodLinks
IPO Entropic τ=10tau (τ)=10, objective=Entropic2026.0553.6
IPO k = 1 Neutralk=1, objective=Neutral2026.0550.8
IPO CVaR α=0.25alpha (α)=0.25, objective=CVaR2026.0549.4
IPO Entropic τ=5tau (τ)=5, objective=Entropic2026.0548.8
IPO CVaR α=0.125alpha (α)=0.125, objective=CVaR2026.0548.6
EG Entropic τ=5tau (τ)=5, objective=Entropic2026.0544.5
EG k = 1 Neutralk=1, objective=Neutral2026.0541.2
NMD k = 1 Neutralk=1, objective=Neutral2026.0539.2
EGPO† (HF)source=Hugging Face2026.059.1
SFT Basetype=Baseline2026.052.4