List your company
ResearchFormerly WizWand

Direct Preference Optimization on SHP AlpacaEval 2.0

18.44LCWRFull Set
1.93526.220110.50514.7899Aug 6, 2025
Updated 5mo ago

Evaluation results

MethodLinks
Full SetSelection Method=Full Set2025.0818.4416.82
Difficulty-Based Preference Data SelectionSelection Method=Ours2025.0818.4119.52
Difficulty-Based Preference Data SelectionSelection Method=Ours2025.0817.9216.74
Full SetSelection Method=Full Set2025.0817.8416.52
ZIP†Selection Method=ZIP2025.0817.8316.77
RandomSelection Method=Random2025.0817.5317.49
ZIP†Selection Method=ZIP2025.0817.2216.03
DiverseEvol†Selection Method=DiverseEvol2025.0817.216.59
DiverseEvol†Selection Method=DiverseEvol2025.0816.9815.77
SDPOSelection Method=SDPO2025.0816.6915.74
RandomSelection Method=Random2025.0816.5815.49
SDPOSelection Method=SDPO2025.0816.5815.96
Tulu3-SFTSelection Method=SFT Baseline2025.082.572.16
Tulu3-SFTSelection Method=SFT Baseline2025.082.572.16