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

Preference Alignment on UFB

83.2Win Rate

CW-DPO

54.18461.71769.2576.783Mar 5, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.03
83.2
2026.03
79.8
2026.03
79.5
2026.03
77.1
2026.03
76.8
2026.03
76.6
2026.03
75.4
2026.03
74.3
2026.03
70.9
2026.03
64.7
2026.03
64.3
2026.03
63.9
2026.03
62.1
2026.03
60.2
2026.03
59.8
2026.03
59.4
2026.03
56.8
2026.03
55.3