List your company

Generative Performance on HH Golden 42.5k (test)

88.9Win RateFA-DPO
18.33636.655554.97573.2945Nov 30, 2025

Evaluation results

MethodLinks
FA-DPOFlipping ratio (η)=10%, Model Backbone=Pythia-1B2025.1188.9
FA-DPOFlipping ratio (η)=0%, Model Backbone=Pythia-1B2025.1188.7
cDPOFlipping ratio (η)=0%, Model Backbone=Pythia-1B2025.1188.35
FA-DPOFlipping ratio (η)=30%, Model Backbone=Pythia-1B2025.1187.7
DPOFlipping ratio (η)=0%, Model Backbone=Pythia-1B2025.1187
FA-DPOFlipping ratio (η)=20%, Model Backbone=Pythia-1B2025.1182.85
rDPOFlipping ratio (η)=10%, Model Backbone=Pythia-1B2025.1182.8
ROPOFlipping ratio (η)=10%, Model Backbone=Pythia-1B2025.1181.15
FA-DPOFlipping ratio (η)=40%, Model Backbone=Pythia-1B2025.1178.6
rDPOFlipping ratio (η)=0%, Model Backbone=Pythia-1B2025.1177.3
ROPOFlipping ratio (η)=0%, Model Backbone=Pythia-1B2025.1170.5
SIMPOFlipping ratio (η)=0%, Model Backbone=Pythia-1B2025.1165.1
cDPOFlipping ratio (η)=20%, Model Backbone=Pythia-1B2025.1162.3
rDPOFlipping ratio (η)=30%, Model Backbone=Pythia-1B2025.1158.1
cDPOFlipping ratio (η)=10%, Model Backbone=Pythia-1B2025.1156.95
ROPOFlipping ratio (η)=20%, Model Backbone=Pythia-1B2025.1154.3
rDPOFlipping ratio (η)=40%, Model Backbone=Pythia-1B2025.1154.1
rDPOFlipping ratio (η)=20%, Model Backbone=Pythia-1B2025.1153.55
ROPOFlipping ratio (η)=30%, Model Backbone=Pythia-1B2025.1153.5
ROPOFlipping ratio (η)=40%, Model Backbone=Pythia-1B2025.1149.2
cDPOFlipping ratio (η)=40%, Model Backbone=Pythia-1B2025.1149.05
cDPOFlipping ratio (η)=30%, Model Backbone=Pythia-1B2025.1137.75
DPOFlipping ratio (η)=10%, Model Backbone=Pythia-1B2025.1134.05
SIMPOFlipping ratio (η)=10%, Model Backbone=Pythia-1B2025.1129.95
DPOFlipping ratio (η)=20%, Model Backbone=Pythia-1B2025.1129.95
SIMPOFlipping ratio (η)=40%, Model Backbone=Pythia-1B2025.1127.5
DPOFlipping ratio (η)=40%, Model Backbone=Pythia-1B2025.1126.85
SIMPOFlipping ratio (η)=20%, Model Backbone=Pythia-1B2025.1124.9
SIMPOFlipping ratio (η)=30%, Model Backbone=Pythia-1B2025.1122.4
DPOFlipping ratio (η)=30%, Model Backbone=Pythia-1B2025.1121.05