Efficient Diffusion Policies for Offline Reinforcement Learning

Offline reinforcement learning (RL) aims to learn optimal policies from offline datasets, where the parameterization of policies is crucial but often overlooked.