PowerSGD: Practical Low-Rank Gradient Compression for Distributed Optimization

05/31/2019
by   Thijs Vogels, et al.
0

We study gradient compression methods to alleviate the communication bottleneck in data-parallel distributed optimization. Despite the significant attention received, current compression schemes either do not scale well or fail to achieve the target test accuracy. We propose a new low-rank gradient compressor based on power iteration that can i) compress gradients rapidly, ii) efficiently aggregate the compressed gradients using all-reduce, and iii) achieve test performance on par with SGD. The proposed algorithm is the only method evaluated that achieves consistent wall-clock speedups when benchmarked against regular SGD with an optimized communication backend. We demonstrate reduced training times for convolutional networks as well as LSTMs on common datasets. Our code is available at https://github.com/epfml/powersgd.

READ FULL TEXT
research
09/26/2021

Quantization for Distributed Optimization

Massive amounts of data have led to the training of large-scale machine ...
research
06/15/2023

Evaluation and Optimization of Gradient Compression for Distributed Deep Learning

To accelerate distributed training, many gradient compression methods ha...
research
08/04/2020

PowerGossip: Practical Low-Rank Communication Compression in Decentralized Deep Learning

Lossy gradient compression has become a practical tool to overcome the c...
research
06/09/2023

Error Feedback Can Accurately Compress Preconditioners

Leveraging second-order information at the scale of deep networks is one...
research
11/08/2018

GradiVeQ: Vector Quantization for Bandwidth-Efficient Gradient Aggregation in Distributed CNN Training

Data parallelism can boost the training speed of convolutional neural ne...
research
10/29/2020

Accordion: Adaptive Gradient Communication via Critical Learning Regime Identification

Distributed model training suffers from communication bottlenecks due to...
research
01/24/2019

Trajectory Normalized Gradients for Distributed Optimization

Recently, researchers proposed various low-precision gradient compressio...

Please sign up or login with your details

Forgot password? Click here to reset