Fix seeding of new generator for multi GPU #3459
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
4 tasks
|
Yes the same should be done there. Do you want to include it in this PR? |
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
Thanks for the quick check. Yes I will add it in this PR. I'll try to add a test as well. |
Done. |
|
The added test fails on main, passes on this branch. Not sure this covers the other line I just changed in the code. |
muellerzr
approved these changes
Mar 28, 2025
|
Thanks for the quick review! |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What does this PR do?
Fixes #3398
The initial state of the new generator was always the same, whatever the seed passed to
accelerate.utils.set_seed. This PR makes sure that we get reproducible results but different numbers for different seeds passed toset_seed. This is already the case when multi GPU is disabled because generator is None (see https://github.com/pytorch/pytorch/blob/main/torch/utils/data/sampler.py#L170).Before adding a test, I will wait for a first reviewer to confirm this is OK, I am not yet very familiar with accelerate.
Maybe the same should be done for https://github.com/huggingface/accelerate/blob/main/src/accelerate/data_loader.py#L1184 where the seed is set to 42 and thus will always produce the same number whatever the first global initial seed passed to
set_seedBefore submitting
Pull Request section?
to it if that's the case.
documentation guidelines, and
here are tips on formatting docstrings.
Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
Based on git blame I would say @muellerzr :)