Provably (More) Sample-Efficient Offline RL with Options

Planning in long-horizon tasks is challenging in reinforcement learning (RL) (Co-Reyes et al., 2018;