Unintentional Unalignment: Likelihood Displacement in Direct Preference Optimization.
Noam Razin, Sadhika Malladi, Adithya Bhaskar, Danqi Chen, Sanjeev Arora, Boris Hanin
Browse the full ICLR paper archive.
Noam Razin, Sadhika Malladi, Adithya Bhaskar, Danqi Chen, Sanjeev Arora, Boris Hanin
Browse the full ICLR paper archive.