Thank you for submitting your work to NeurIPS. All reviewers were enthusiastic about the paper, and I am happy to accept it. Strong empirical results and practicality of the method were appreciated by the reviewers. Expert reviewers found the method sufficiently novel for publication. However, the fact that other papers use the exponential moving average (e.g. http://proceedings.mlr.press/v80/jiang18c/jiang18c.pdf) has to be discussed much more clearly. Please make sure that you include in the introduction (not just related work) section a detailed discussion on prior work that was flagged by R1. Please also make sure to include a full comparison with prior work (as in Table 1 in the rebuttal).