Skip to yearly menu bar Skip to main content


Poster

ATOMO: Communication-efficient Learning via Atomic Sparsification

Hongyi Wang · Scott Sievert · Shengchao Liu · Zachary Charles · Dimitris Papailiopoulos · Stephen Wright

Room 210 #9

Keywords: [ Optimization ] [ Optimization for Deep Networks ] [ Communication- or Memory-Bounded Learning ]


Abstract:

Distributed model training suffers from communication overheads due to frequent gradient updates transmitted between compute nodes. To mitigate these overheads, several studies propose the use of sparsified stochastic gradients. We argue that these are facets of a general sparsification method that can operate on any possible atomic decomposition. Notable examples include element-wise, singular value, and Fourier decompositions. We present ATOMO, a general framework for atomic sparsification of stochastic gradients. Given a gradient, an atomic decomposition, and a sparsity budget, ATOMO gives a random unbiased sparsification of the atoms minimizing variance. We show that recent methods such as QSGD and TernGrad are special cases of ATOMO, and that sparsifiying the singular value decomposition of neural networks gradients, rather than their coordinates, can lead to significantly faster distributed training.

Live content is unavailable. Log in and register to view live content