HiL-ResRL: A Model-Agnostic Finetuning Adapter via Human-in-the-loop Residual Reinforcement Learning figure
AlphaXiv 中文概览(可滚动查看)