An Efficient Statistical-based Gradient Compression Technique for Distributed Training Systems
This work addresses communication efficiency for distributed deep learning systems, offering an incremental improvement over existing compression methods.
The paper tackles the challenge of communication bottlenecks in distributed training of large neural networks by proposing SIDCo, a gradient compression technique that models gradients with sparsity-inducing distributions, achieving speedups of up to 41.7% over no compression and 1.9% over DGC.
The recent many-fold increase in the size of deep neural networks makes efficient distributed training challenging. Many proposals exploit the compressibility of the gradients and propose lossy compression techniques to speed up the communication stage of distributed training. Nevertheless, compression comes at the cost of reduced model quality and extra computation overhead. In this work, we design an efficient compressor with minimal overhead. Noting the sparsity of the gradients, we propose to model the gradients as random variables distributed according to some sparsity-inducing distributions (SIDs). We empirically validate our assumption by studying the statistical characteristics of the evolution of gradient vectors over the training process. We then propose Sparsity-Inducing Distribution-based Compression (SIDCo), a threshold-based sparsification scheme that enjoys similar threshold estimation quality to deep gradient compression (DGC) while being faster by imposing lower compression overhead. Our extensive evaluation of popular machine learning benchmarks involving both recurrent neural network (RNN) and convolution neural network (CNN) models shows that SIDCo speeds up training by up to 41:7%, 7:6%, and 1:9% compared to the no-compression baseline, Topk, and DGC compressors, respectively.