Excited to share that our paper “Provably Robust DPO: Aligning Language Models with Noisy Feedback” has been accepted at #ICML2024!
We introduce Robust DPO, an unbiased estimate of the DPO loss that is robust to preference noise in the data.
arxiv.org/abs/2403.00409
🧵[1/n]
I request the embassy to give some priority slots for J1 candidates. We are comparatively less in number and being overshadowed by F1 applicants. It's a very scary situation. Unable to book slots from past one month. @POTUS@USAndIndia@USAndMumbai