The reviewers agree that this paper makes a solid contribution to the literature of differentially private synthetic data. In particular, the paper establishes new connection with private PAC learning for new regimes. The authors should follow reviewers' suggestions to revise the paper and consider citing suggested related work.