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

Human Preference Alignment on ArenaHard V2

60Avg@3 Score

Qwen3-30A3-2507

25.05634.12843.252.272Dec 6, 2025
Updated 1mo ago

Evaluation Results

MethodLinks
2025.12
60
2025.12
60
2025.12
48.4
2025.12
40.5
2025.12
39.9
2025.12
26.4