fix comment misdirection during scaling loss #36987
Open
+1
−1
Conversation
This file contains 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
|
Hi 👋, thank you for opening this pull request! The pull request is converted to draft by default. The CI will be paused while the PR is in draft mode. When it is ready for review, please click the |
SunMarc
reviewed
Mar 26, 2025
Comment on lines
+3755
to
3757
| # Finally we need to normalize the loss if GA loss bug is not fixed during compute loss | ||
| if not self.model_accepts_loss_kwargs and self.compute_loss_func is None: | ||
| loss = loss / self.args.gradient_accumulation_steps |
There was a problem hiding this comment.
I feel like this condition if not strong enough. Users can overwrite compute loss and not use num_item_per_batch at all when calculating the loss (e.g trl case as I explained here ). The issue is that right now it won't scale the loss.
Maybe we should create a variable to enable/disable the GA fix but in transformers, we set enable it since we are calculating the loss the right way.
SunMarc
reviewed
Mar 26, 2025
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Outdated suggestions cannot be applied.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?
If the GA loss bug is not fixed, we should scale the loss properly. However, it is not solely for reporting purposes as described in the previous comment. https://github.com/huggingface/transformers/blob/main/src/transformers/trainer.py#L3764
Before submitting
Pull Request section?
to it if that's the case.
documentation guidelines, and
here are tips on formatting docstrings.
Who can review?
@SunMarc @muellerzr @ArthurZucker