Toward Efficient Low-Precision Training: Data Format Optimization and Hysteresis Quantization

Abstract

As the complexity and size of deep neural networks continue to increase, low-precision training has been extensively studied in the last few years to reduce hardware overhead. Training performance is largely affected by the numeric formats representing different values in low-precision training, but finding an optimal format typically requires numerous training runs, which is a very time-consuming process. In this paper, we propose a method to efficiently find an optimal format for activations and errors without actual training. We employ this method to determine an 8-bit format suitable for training various models. In addition, we propose hysteresis quantization to suppress undesired fluctuation in quantized weights during training. This scheme enables deeply quantized training using 4-bit weights, exhibiting only 0.2% degradation for ResNet-18 trained on ImageNet.

Cite

Text

Lee et al. "Toward Efficient Low-Precision Training: Data Format Optimization and Hysteresis Quantization." International Conference on Learning Representations, 2022.

Markdown

[Lee et al. "Toward Efficient Low-Precision Training: Data Format Optimization and Hysteresis Quantization." International Conference on Learning Representations, 2022.](https://mlanthology.org/iclr/2022/lee2022iclr-efficient/)

BibTeX

@inproceedings{lee2022iclr-efficient,
  title     = {{Toward Efficient Low-Precision Training: Data Format Optimization and Hysteresis Quantization}},
  author    = {Lee, Sunwoo and Park, Jeongwoo and Jeon, Dongsuk},
  booktitle = {International Conference on Learning Representations},
  year      = {2022},
  url       = {https://mlanthology.org/iclr/2022/lee2022iclr-efficient/}
}