Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Sync b/w Eleuther AI's DeeperSpeed and DeepSpeed #987

Closed
g-karthik opened this issue Apr 21, 2021 · 3 comments
Closed

Sync b/w Eleuther AI's DeeperSpeed and DeepSpeed #987

g-karthik opened this issue Apr 21, 2021 · 3 comments

Comments

@g-karthik
Copy link

Can the added changes made in Eleuther AI's DeeperSpeed fork be merged back into DeepSpeed?

I see some critical differences, such as in PipelineModule: former and latter.

When using containerized setups, it's a lot cleaner to just install from one source rather than have multiple things going on.

@tjruwase @jeffra @ShadenSmith

@sdtblck
Copy link
Contributor

sdtblck commented Apr 21, 2021

I submitted a PR for the critical difference you linked above a while back but I think it went unnoticed. #687

It's been paramount for us to get Pipeline Parallelism working at all, so I'm surprised it's still in the main branch.

@sdtblck
Copy link
Contributor

sdtblck commented Apr 21, 2021

Aside from that and this EleutherAI@0e95737 (for running inference - majorly hacky and I'm sure there's a better way of doing it), they're mostly the same aside from some minor changes to timers / logging.

@loadams
Copy link
Contributor

loadams commented Aug 18, 2023

Closing as this is stale and not an active issue. If there are other changes from a fork people want merged, please link this when you make a PR.

@loadams loadams closed this as completed Aug 18, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants