NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:1688
Title:On Lazy Training in Differentiable Programming

The paper received no strongly positive scores 6/5/6 at the beginning, with acceptable confidences 4/4/3. The author rebuttal successfully convinced reviewers #1 and #2, who both raised their scores by 1. But reviewer #3 was still not fully convinced so s/he maintained her/his score. So the new scores are 7/6/6. The AC browsed through the paper and rebuttal, and found that the paper does provide good insight into the performance of DNNs, although the DNN models assumed are still far from those in real use. So the AC recommended accepting as poster.