The authors propose a deep learning method to generate molecular structures that are synthesizable and give a recipe for their synthesis from reactant molecules. The reviewers fond the contribution be significant for practioners. The authors addresed the unclear points in their response.