Task Specialization Fine-Tuning for Contextual Reinforcement Learning

Jianan Zhou, Jung-Hoon Cho, Tianyue Zhou +5

Contextual Reinforcement Learning (CRL) seeks to generalize classical RL by maximizing task coverage across a context space of related tasks. While prior works often train from scr…