Review for NeurIPS paper: BOSS: Bayesian Optimization over String Spaces

Additional Feedback: (1) I am curious about the differences between the proposed work and this one: Amortized Bayesian Optimization over Discrete Spaces, by Yulia Rubanov, etc. It seems the general idea is quite similar, for example, they also employ the evolutionary algorithms in BO. (2) What's the maximized length of the string did you test in the experiment? Besides the GA, I am curious if the authors have tried some other methods, such as random forest. In your scenario, have you ever considered using different kernels to build the objective function? Obviously, this work will be welcomed in the community. By considering the original good work and the satisfied feedback, I'd like to rise my score to "A good submission, accept".