List your company

Dialogue Generation on Anthropic-HH (test)

69.07Average Preference ScoreCal-DPO
42.997249.766156.53563.3039Dec 19, 2024Feb 26, 2025May 7, 2025Jul 15, 2025Sep 23, 2025Dec 1, 2025Feb 9, 2026

Evaluation results

MethodLinks
Cal-DPO2024.1269.0773.5264.61
DDPOTemperature=0.72026.0268.72——
DPO*Temperature=0.72026.0268.52——
DPO*Temperature=0.12026.0268.02——
CPO2024.1267.8672.363.39
DDPOTemperature=0.12026.0267.05——
DPO2024.1264.3469.3259.35
DPOP2024.1263.3468.7757.91
SLiC2024.1261.6265.5257.71
DPO+NLL2024.1260.3165.3455.28
IPO2024.1259.1663.1955.12
DPOTemperature=0.72026.0258.99——
f-DPO2024.1256.360.2152.38
DPOTemperature=0.12026.0255.3——
SFTTemperature=0.72026.0249.48——
SFTTemperature=0.12026.0244——