SimPO: Simple Preference Optimization with a Reference-Free Reward

Devs

SimPO: Simple Preference Optimization with a Reference-Free Reward | Read Paper on Bytez