Published April 10, 2021 | Version v1

SC21 AD/AE for "Why Globally Re-shuffle? An I/O Perspective on Data Shuffling in Large Scale Deep Learning"

Authors/Creators

  • 1. AIST, Japan

Description

We use ABCI (https://abci.ai/) for testing the effect of shuffling techniques on convergence of distributed SGD training
This repository contains the artifact, e.g. job scripts, models, datasets.

Files

Shuffling-SC21_AD.zip

Files (231.7 kB)

Name Size Download all
md5:1f6497f934792f2d116dcf61a7ccf496
231.7 kB Preview Download