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

Update mosaic_fsdp_utils.py #3185

Merged
merged 1 commit into from
Apr 10, 2024
Merged

Update mosaic_fsdp_utils.py #3185

merged 1 commit into from
Apr 10, 2024

Conversation

vchiley
Copy link
Contributor

@vchiley vchiley commented Apr 10, 2024

Porting mosaicml/llm-foundry#1104 to composer.

What does this PR do?

Updates how a new process group is initialized. This version was broken in some version of torch1.XX, but has been working well in torch2+

What issue(s) does this change relate to?

Before submitting

  • Have you read the contributor guidelines?
  • Is this change a documentation change or typo fix? If so, skip the rest of this checklist.
  • Was this change discussed/approved in a GitHub issue first? It is much more likely to be merged if so.
  • Did you update any related docs and document your change?
  • Did you update any related tests and add any new tests related to your change? (see testing)
  • Did you run the tests locally to make sure they pass?
  • Did you run pre-commit on your change? (see the pre-commit section of prerequisites)

Copy link
Contributor

@mvpatel2000 mvpatel2000 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM! Assuming you tested?

@vchiley
Copy link
Contributor Author

vchiley commented Apr 10, 2024

LGTM! Assuming you tested?

Not yet.
Theres a reason this is a draft 😅

@vchiley
Copy link
Contributor Author

vchiley commented Apr 10, 2024

(I set really short warmups which explains some of the "loss spikes" that come back down)

Screenshot 2024-04-10 at 11 27 03 AM

from here
The newest version is highlighted in aqua
Pink is the true baseline
Light blue is the baseline before mosaicml/llm-foundry#1104

tldr lgtm
although I have no clue what the failing test is...

@vchiley vchiley marked this pull request as ready for review April 10, 2024 18:30
@vchiley vchiley merged commit 2a262b4 into mosaicml:dev Apr 10, 2024
14 checks passed
vchiley added a commit that referenced this pull request Apr 12, 2024
vchiley added a commit that referenced this pull request Apr 12, 2024
staghado pushed a commit to lightonai/composer that referenced this pull request Apr 13, 2024
staghado pushed a commit to lightonai/composer that referenced this pull request Apr 13, 2024
DhruvDh pushed a commit to DhruvDh/composer that referenced this pull request Apr 21, 2024
j316chuck pushed a commit that referenced this pull request May 16, 2024
j316chuck pushed a commit that referenced this pull request May 16, 2024
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

Successfully merging this pull request may close these issues.

2 participants