ResearchBenchmarksPreference Learning on Simulated Matchmaking Environment (p_flip=0.2, T=6,400)Follow80Like RateTK76788082Apr 11, 2026Evaluation ResultsMethodMethodLinksLike RateAlign@20Delta S (Δs)TKα=1α=12026.04807194.8Block-TKα=1α=12026.048069.896.8Block-NK2026.048069.899.4NKη=0.5η=0.52026.048062.673.8K-NoNorm2026.04807194.8OGD-0.1parameter=0.1parameter=0.12026.048071.693.3