2) Showed how CANs and the SL follow naturally from introducing constraints into the discriminator of GANs; 3)

We thank all the reviewers for their careful reviews. None of these follow trivially from Xu et al. In molecule generation, validity is encouraged using the reward-based approach of MolGANs, not the SL. As a result diversity increases (91% to 98.98%) and so does uniqueness (2.4 to 2.5). We will clarify this in the discussion.