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

Preference Optimization on Roleplay 1500 users

90.9Winrate

Oracle

48.36459.40770.4581.493Feb 26, 2025
Updated 1mo ago

Evaluation Results

MethodLinks
2025.02
90.9
2025.02
90.3
2025.02
82.6
2025.02
81.3
2025.02
72.4
2025.02
67.3
2025.02
63.2
2025.02
62.8
2025.02
53.3
50