Skip to content

Conversation

@muellerzr
Copy link
Contributor

@muellerzr muellerzr commented Oct 16, 2023

What does this PR do?

This PR brings back the set_epoch logic and solves the last remaining regression from the accelerate integration. Resuming training should now be exactly identical. (Introduced due to an incomplete implementation in #24028)

Linked with huggingface/accelerate#2057

Fixes # (issue)

fixes #26541

Before submitting

  • This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
  • Did you read the contributor guideline,
    Pull Request section?
  • Was this discussed/approved via a Github issue or the forum? Please add a link
    to it if that's the case.
  • Did you make sure to update the documentation with your changes? Here are the
    documentation guidelines, and
    here are tips on formatting docstrings.
  • Did you write any new necessary tests?

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.

@LysandreJik @ArthurZucker

@HuggingFaceDocBuilderDev
Copy link

HuggingFaceDocBuilderDev commented Oct 16, 2023

The documentation is not available anymore as the PR was closed or merged.

Copy link
Collaborator

@ArthurZucker ArthurZucker left a comment

Choose a reason for hiding this comment

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

The tests seem to error out, so waiting for this.
Could you link to which pr removed it?

@LysandreJik LysandreJik merged commit 9041240 into main Oct 26, 2023
@LysandreJik LysandreJik deleted the muellerzr-dataloader branch October 26, 2023 09:20
EduardoPach pushed a commit to EduardoPach/transformers that referenced this pull request Nov 19, 2023
…26850)

* Working tests!

* Fix sampler

* Fix

* Update src/transformers/trainer.py

Co-authored-by: Arthur <[email protected]>

* Fix check

* Clean

---------

Co-authored-by: Arthur <[email protected]>
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.

Using IterableDataset, Trainer never calls set_epoch to increment and resets the epochs to 0 at the beginning of each epoch.

5 participants