arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2002.11082cs.LGcs.DCstat.ML
收藏

通信高效分布式训练的最优梯度Quantization条件

Optimal Gradient Quantization Condition for Communication-Efficient Distributed Training

University of Pittsburgh(匹兹堡大学)

An Xu, Zhouyuan Huo, Heng Huang

摘要:

在计算机视觉应用中,梯度的通信对于用多个设备训练深度神经网络是昂贵的。特别是,深度学习模型的规模不断增长导致更高的通信开销,这违背了关于设备数量的理想线性训练加速。梯度量化是降低通信开销的常用方法之一。然而,它会导致量化误差的训练和模型性能下降。本文推导了\textbf{ANY}梯度分布的二值和多值梯度量化的最佳条件。基于最优条件,我们提出了两种新的量化方案:有偏BinGrad和无偏ORQ,分别用于二值和多级梯度量化,动态确定最优量化级别。在CIFAR和ImageNet数据集上使用几种流行的卷积神经网络进行的实验结果表明了我们所提出的方法的优越性。

英文摘要:

The communication of gradients is costly for training deep neural networks with multiple devices in computer vision applications. In particular, the growing size of deep learning models leads to higher communication overheads that defy the ideal linear training speedup regarding the number of devices. Gradient quantization is one of the common methods to reduce communication costs. However, it can lead to quantization error in the training and result in model performance degradation. In this work, we deduce the optimal condition of both the binary and multi-level gradient quantization for \textbf{ANY} gradient distribution. Based on the optimal condition, we develop two novel quantization schemes: biased BinGrad and unbiased ORQ for binary and multi-level gradient quantization respectively, which dynamically determine the optimal quantization levels. Extensive experimental results on CIFAR and ImageNet datasets with several popular convolutional neural networks show the superiority of our proposed methods.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑