Clarify that causal LM labels are shifted internally#44642
Open
joshuaswanson wants to merge 1 commit intohuggingface:mainfrom
Open
Clarify that causal LM labels are shifted internally#44642joshuaswanson wants to merge 1 commit intohuggingface:mainfrom
joshuaswanson wants to merge 1 commit intohuggingface:mainfrom
Conversation
Contributor
|
View the CircleCI Test Summary for this PR: https://huggingface.co/spaces/transformers-community/circle-ci-viz?pr=44642&sha=1cb1e5 |
This was referenced Apr 29, 2026
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The generic
labelsdocstring inModelArgssays "masked language modeling loss" and doesn't mention that causal LM models shift labels internally. This has tripped up a lot of users who pre-shift their labels and end up training next-next-token prediction by accident.Updates the shared docstring to say "language modeling loss" (since it's used by causal LM models too, not just masked LM) and adds a note explaining that causal LM models handle the shifting, so users should pass
labels = input_ids.Fixes #32944