Skip to content

srikarchundury/horovodContainerRingAlgo

BranchesTags

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
.ipynb_checkpoints
.ipynb_checkpoints
 
 
arm_image
arm_image
 
 
comp
comp
 
 
docker
docker
 
 
old_results
old_results
 
 
results
results
 
 
6_workers.png
6_workers.png
 
 
Abstract_Submission.pdf
Abstract_Submission.pdf
 
 
README.md
README.md
 
 
Srikar-Present.pptx
Srikar-Present.pptx
 
 
cluster-compose.yml
cluster-compose.yml
 
 
driver-compose.yml
driver-compose.yml
 
 
exec_into_driver.sh
exec_into_driver.sh
 
 
final_report.pdf
final_report.pdf
 
 
image_prune.sh
image_prune.sh
 
 
measure.sh
measure.sh
 
 
network_6_workers.png
network_6_workers.png
 
 
record_docker_stats.sh
record_docker_stats.sh
 
 
remove_dangling_images.sh
remove_dangling_images.sh
 
 
run_cluster.sh
run_cluster.sh
 
 
run_driver.sh
run_driver.sh
 
 
run_program_in_driver.sh
run_program_in_driver.sh
 
 
statistics-latest.ipynb
statistics-latest.ipynb
 
 
statistics.ipynb
statistics.ipynb
 
 
updated_network_image.png
updated_network_image.png
 
 

Repository files navigation

Step-wise Averaging version of Ring All Reduce Algorithms in Distributed Deep Learning

Attached are the results and my final report here.

The below image captures the jist of my idea. This way we could reduce the network activity for faster distributed training across geographically far-away HPC/AI systems.
main_idea

About

Distributed Deep Learning using Uber's Horovod

Resources

Readme
Activity

Stars

0 stars

Watchers

1 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

No packages published

Languages