This blog summarizes our work of error bounds of imitating policies and environments, which is presented at NeurIPS 2020.