AI RESEARCH

The Representation-Rationalizability Tradeoff in Reward Learning

arXiv CS.LG

ArXi:2606.00291v1 Announce Type: cross In RLHF, each