Deep Double Descent: Where Bigger Models and More Data Hurt

Preetum Nakkiran, Gal Kaplun, Yamini Bansal, Tristan Yang, Boaz Barak, Ilya Sutskever

TL;DR: We demonstrate, and characterize, realistic settings where bigger models are worse, and more data hurts.

Abstract: We show that a variety of modern deep learning tasks exhibit a "double-descent" phenomenon where, as we increase model size, performance first gets worse and then gets better. Moreover, we show that double descent occurs not just as a function of model size, but also as a function of the number of training epochs. We unify the above phenomena by defining a new complexity measure we call the effective model complexity, and conjecture a generalized double descent with respect to this measure. Furthermore, our notion of model complexity allows us to identify certain regimes where increasing (even quadrupling) the number of train samples actually hurts test performance.

Keywords: deep learning, double descent, optimization, SGD, complexity

Code: [ 3 community implementations](https://paperswithcode.com/paper/?openreview=B1g5sA4twr)

Data: [CIFAR-10](https://paperswithcode.com/dataset/cifar-10), [CIFAR-100](https://paperswithcode.com/dataset/cifar-100)

Community Implementations: [ 6 code implementations](https://www.catalyzex.com/paper/arxiv:1912.02292/code)