[VisionEncoderDecoderModel] Update loss function#40863
Merged
Merged
Conversation
|
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. |
Contributor
|
[For maintainers] Suggested jobs to run (before merge) run-slow: vision_encoder_decoder |
Collaborator
Author
|
cc @ArthurZucker this one can be merged |
ngazagna-qc
pushed a commit
to ngazagna-qc/transformers
that referenced
this pull request
Oct 23, 2025
Update loss function
4 tasks
SangbumChoi
pushed a commit
to SangbumChoi/transformers
that referenced
this pull request
Jan 23, 2026
Update loss function
pull Bot
pushed a commit
to dwongdev/transformers
that referenced
this pull request
Jun 24, 2026
…abels[..., 1:]) (huggingface#46784) * Fix Moonshine training-loss double-shift Moonshine right-shifts labels into decoder_input_ids, then computes the loss via ForCausalLMLoss, which shifts again, so it trains against labels[..., 1:]. Use a plain CrossEntropyLoss instead, matching Whisper/Bart and the VisionEncoderDecoder fix in huggingface#40863. Adds a regression test (with and without -100 padding). * Regenerate Moonshine streaming model * Mirror the training-loss regression test in moonshine_streaming make fix-repo propagated the loss fix to the streaming model, so the same no-double-shift test should guard it there too.
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
3 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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?
Models like Donut are currently broken on main, they can't be fine-tuned. In order to unblock users at #39473, this PR reverts #36753.
It looks like the
ForCausalLMLossclass shifts the labels, however the VisionEncoderDecoderModel class does not expect shifted labels as seen here.