Optimizing Language Models for Human Preferences is a Causal Inference Problem.

Victoria Lin, Eli Ben-Michael, Louis-Philippe Morency

Browse the full UAI paper archive.