🎭 Fix train and eval mode checking in GRPOTrainer and SFTTrainer
#3337
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
qgallouedec
approved these changes
Apr 25, 2025
There was a problem hiding this comment.
That's a very good point @I-l-l-I!
I've tested myself I can confirm that the current implementation has indeed the issue you describe, and that your solution works like a charm.
I've also fixed it in SFT.
GRPOTrainer and SFTTrainer
|
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. |
shirinyamani
pushed a commit
that referenced
this pull request
Apr 28, 2025
…3337) Co-authored-by: Jiaming Ma <jiaming.ma@connect.polyu.hk> Co-authored-by: Quentin Gallouédec <45557362+qgallouedec@users.noreply.github.com> Co-authored-by: Quentin Gallouédec <gallouedec.quentin@gmail.com>
4 tasks
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Outdated suggestions cannot be applied.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?
Use
model.traininginstead ofcontrol.should_evaluateto determine if the model is in evaluation mode.Motivation
It is not accurate to use
should_evaluateto determine themode. For example, if the trainer needs to evaluate and log at a certain training step, thenlogwill be executed twice at this step. The first log should be for the training metrics, but at this timeshould_evaluate=True, sinceshould_evaluatemeans to evaluate after one step.Before submitting
Pull Request section?
to it if that's the case.
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.