Provably Robust DPO: Aligning Language Models with Noisy Feedback.

Sayak Ray Chowdhury, Anush Kini, Nagarajan Natarajan

Browse the full ICML paper archive.