Provably avoiding over-optimization in Direct Preference Optimization without knowing the data distribution

Add code
Feb 05, 2026

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: