Reviews: Scaling the Poisson GLM to massive neural datasets through polynomial approximations

Given that most of the method is derived from the ideas in the PASS-GLM paper, I think (a) and (b) are less significant results. Moreover, it is hard to judge the significance of the extensions to the method (c). For example, the authors did not report compare their results on using the adaptive intervals in section 4.2 to using fixed intervals, instead simply stating that they "find that this procedure provides large improvements" (line 185).