The reviewers agree that the techniques leveraged in the paper should be of interest to the wider NeurIPS community. Furthermore, even though the setting analyzed is relatively simple, the analysis is challenging, and understanding the effects of batch size is a problem of broad interest.