Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Direct Preference Optimization on Skywork AlpacaEval 2.0

20.56LCWR

Difficulty-Based Preference Data Selection

1.85046.707711.56516.4223Aug 6, 2025
Updated 16d ago

Evaluation Results

MethodLinks
2025.08
20.5619.38
2025.08
19.618.57
2025.08
18.7418.96
2025.08
18.1317.54
2025.08
17.7518.33
2025.08
17.4618.56
2025.08
2.572.16