NeurIPS 2020

Pushing the Limits of Narrow Precision Inferencing at Cloud Scale with Microsoft Floating Point

Meta Review

Four knowledgeable referees support acceptance for the contribution mainly due to its novelty and practical impact, and I also recommend acceptance. The only concern is about clarity and reproducibility; Most of the uncertain parts raised by reviewers have been resolved through rebuttal, but I ask authors to thoroughly go through the paper so that there is no lack of explanation in all areas of the final version.