Skip to content

Code for the paper "Linearly Converging Error Compensated SGD"

Notifications You must be signed in to change notification settings

eduardgorbunov/ef_sigma_k

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Linearly Converging Error Compensated SGD

Code for the paper "Linearly Converging Error Compensated SGD" by Eduard Gorbunov, Dmitry Kovalev, Dmitry Makarenko and Peter Richtárik, NeurIPS 2020.

To run the experiments one should download the data from LIBSVM https://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/binary.html and save the datasets in .txt format in the folder called "datasets". Also one should create folders "dump" and "plot".

Use jupyter notebooks that we prepared in order to repeat/check our experiments.

distrib_algs.py -- file with implementation of methods

utils.py -- file with useful tools for saving results and making plots

functions.py -- file with implementation of oracles and compression/quantization operators that we use

About

Code for the paper "Linearly Converging Error Compensated SGD"

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published