SimPO: Simple Preference Optimization with a Reference-Free Reward
Primary LanguagePythonMIT LicenseMIT