$i$REPO: $i$mplicit Reward Pairwise Difference based Empirical Preference Optimization

Open in new window