Sayak Ray Chowdhury

Papers on Lattice

Total citations

Topics

h-index

Research focus

RLHF & Preference Learning (1)Scalable Oversight & Alignment Theory (1)

Frequent co-authors

Aditya Gopalan (1)Debangshu Banerjee (1)

Papers (1)

Oct 23, 2025

Aditya Gopalan +2Oct 23, 2025

Why DPO is a Misspecified Estimator and How to Fix It

DPO, a popular RLHF alternative, can actually *hurt* performance due to statistical misspecification, but a simple fix (AuxDPO) can bring it back on track.

Aditya Gopalan, Sayak Ray Chowdhury, Debangshu Banerjee

RLHF & Preference Learning Scalable Oversight & Alignment Theory

Search

Sayak Ray Chowdhury

Research focus

Frequent co-authors

Papers (1)