Adaptive Loss Scaling for Mixed Precision Training

Ruizhe Zhao, Brian Vogel, Tanvir Ahmed

TL;DR: We devise adaptive loss scaling to improve mixed precision training that surpass the state-of-the-art results.

Abstract: Mixed precision training (MPT) is becoming a practical technique to improve the speed and energy efficiency of training deep neural networks by leveraging the fast hardware support for IEEE half-precision floating point that is available in existing GPUs. MPT is typically used in combination with a technique called loss scaling, that works by scaling up the loss value up before the start of backpropagation in order to minimize the impact of numerical underflow on training. Unfortunately, existing methods make this loss scale value a hyperparameter that needs to be tuned per-model, and a single scale cannot be adapted to different layers at different training stages. We introduce a loss scaling-based training method called adaptive loss scaling that makes MPT easier and more practical to use, by removing the need to tune a model-specific loss scale hyperparameter. We achieve this by introducing layer-wise loss scale values which are automatically computed during training to deal with underflow more effectively than existing methods. We present experimental results on a variety of networks and tasks that show our approach can shorten the time to convergence and improve accuracy, compared with using the existing state-of-the-art MPT and single-precision floating point.

Code: https://github.com/ada-loss/ada-loss

Keywords: Deep Learning, Mixed Precision Training, Loss Scaling, Backpropagation

Community Implementations: [ 1 code implementation](https://www.catalyzex.com/paper/arxiv:1910.12385/code)