How Far I'll Go: Offline Goal-Conditioned Reinforcement Learning via $ f $-Advantage Regression

Publication
arXiv preprint arXiv:2206.03023