NeurIPS 2020

TinyTL: Reduce Memory, Not Parameters for Efficient On-Device Learning

Meta Review

All four referees support accept. At some angle, AC thinks that novelty might be a bit limited and the proposed scheme is of engineering type. However, as pointed out by reviewers, AC also agree that the problem studied in this paper is important/interesting, and the ideas/results are practical/impressive. Hence, AC recommends acceptance. AC also suggests the authors to clarify the write-ups following the reviewers' suggestions in the final draft.