All the reviewers agreed that the paper is novel, interesting, and worth to be published in NeurIPS. Please take into account the reviewers' comment in preparing the camera-ready version, especially the expansion of the literature on DFO methods.