English

Decoupled Weight Decay Regularization

Machine Learning 2019-01-08 v3 Neural and Evolutionary Computing Optimization and Control

Abstract

L2_2 regularization and weight decay regularization are equivalent for standard stochastic gradient descent (when rescaled by the learning rate), but as we demonstrate this is \emph{not} the case for adaptive gradient algorithms, such as Adam. While common implementations of these algorithms employ L2_2 regularization (often calling it "weight decay" in what may be misleading due to the inequivalence we expose), we propose a simple modification to recover the original formulation of weight decay regularization by \emph{decoupling} the weight decay from the optimization steps taken w.r.t. the loss function. We provide empirical evidence that our proposed modification (i) decouples the optimal choice of weight decay factor from the setting of the learning rate for both standard SGD and Adam and (ii) substantially improves Adam's generalization performance, allowing it to compete with SGD with momentum on image classification datasets (on which it was previously typically outperformed by the latter). Our proposed decoupled weight decay has already been adopted by many researchers, and the community has implemented it in TensorFlow and PyTorch; the complete source code for our experiments is available at https://github.com/loshchil/AdamW-and-SGDW

Keywords

Cite

@article{arxiv.1711.05101,
  title  = {Decoupled Weight Decay Regularization},
  author = {Ilya Loshchilov and Frank Hutter},
  journal= {arXiv preprint arXiv:1711.05101},
  year   = {2019}
}

Comments

Published as a conference paper at ICLR 2019

R2 v1 2026-06-22T22:45:33.152Z