Skip to content

Upgrade to Transformers v4.55#2209

Merged
regisss merged 33 commits into
mainfrom
transformers_future
Sep 3, 2025
Merged

Upgrade to Transformers v4.55#2209
regisss merged 33 commits into
mainfrom
transformers_future

Conversation

@regisss
Copy link
Copy Markdown
Collaborator

@regisss regisss commented Aug 18, 2025 •
edited
Loading

What does this PR do?

Aligning the codebase with Transformers v4.55.2.

Known issues:

  • text-generation tests are failing for a few models

Before submitting

  • This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
  • Did you make sure to update the documentation with your changes?
  • Did you write any new necessary tests?

regisss and others added 30 commits April 18, 2025 08:08
@regisss
Update setup.py and examples to Transformers v4.51
e6980cb
@regisss
Upgrade generation/utils.py to v4.51.3 (except _beam_search, will…
9786545 
… be done later)
@regisss
Upgrade to Transformers v4.51
0cd573e
@regisss
Fixes
34f32a9
@regisss
Merge branch 'main' into transformers_future
94be78d
@regisss
Fix AWQ requirements
25b053c
@regisss
Merge branch 'main' into transformers_future
ef27edc
@regisss
Beam search
071fb63
@regisss
Make style
b4e3ed3
@regisss
Text-generation fix
831bfae
@regisss
1x tests fix
0947f5b
@regisss
Merge branch 'main' into transformers_future
85ec584
@regisss
Fix 8x tests
8b2e896
@regisss
Merge branch 'main' into transformers_future
7ef7282
@regisss
Merge branch 'main' into transformers_future
452b66a
@regisss
Fix 8x tests
762c08f
@regisss
Fix mllama test
1c91d3f
@IlyasMoutawwakil
Merge branch 'main' into transformers_future
4636aef
@regisss
Fix CI tests
31bf07e
@regisss
Merge branch 'main' into transformers_future
3805cea
@regisss
Merge branch 'main' into transformers_future
01dc42d
@regisss
Merge branch 'main' into transformers_future
c0893c2
@regisss
Merge branch 'main' into transformers_future
f0a5945
@regisss
Update examples
3515a12
@regisss
generation + integrations + loss folders
d4960e7
@regisss
models
91e29c4
@regisss
trainer + tests
5b39327
@regisss
Fix tests + comment remote models (to fix later)
5450d7a
@regisss
Fix decilm
f8aa93a
@regisss
Fix deepseek_v2
f1de537
@pbielak
Copy link
Copy Markdown
Collaborator

pbielak commented Aug 19, 2025

@regisss What do you think about moving changes related to:

  1. huggingface-cli login -> hf auth login
  2. type hints
  3. typos, removed comments
    into separate commits/PRs? It would make it easier to review the actual changes coming from the transformers upgrade

If you agree, I could prepare such PRs and then this branch could be rebased

@regisss
Copy link
Copy Markdown
Collaborator Author

regisss commented Aug 19, 2025

@pbielak I don't have a strong opinion on this, go ahead if you prefer reviewing a shorter PR 🙂

@karol-brejna-i karol-brejna-i self-assigned this Aug 20, 2025
This was referenced Aug 21, 2025
Merged
Merged
Merged
Merged
Merged
Merged
@HuggingFaceDocBuilderDev
Copy link
Copy Markdown

HuggingFaceDocBuilderDev commented Aug 21, 2025

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.

pbielak
pbielak reviewed Aug 22, 2025
View reviewed changes
Comment thread optimum/habana/transformers/models/paligemma/modeling_paligemma.py
- add new args token_idx
"""

if (input_ids is None) ^ (inputs_embeds is not None):
Copy link
Copy Markdown
Collaborator

@pbielak pbielak Aug 22, 2025

Choose a reason for hiding this comment

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

Should this be removed?

Copy link
Copy Markdown
Collaborator Author

@regisss regisss Aug 24, 2025

Choose a reason for hiding this comment

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

It's done in PaliGemmaModel: https://github.com/huggingface/transformers/blob/v4.55.4/src/transformers/models/paligemma/modeling_paligemma.py#L324
But I expect this model to fail and to require a fix, I need to run the tests.

Copy link
Copy Markdown
Collaborator

@pbielak pbielak Sep 2, 2025

Choose a reason for hiding this comment

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

So if it's in the upstream, why is it removed here?

Copy link
Copy Markdown
Collaborator Author

@regisss regisss Sep 3, 2025

Choose a reason for hiding this comment

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

What I mean is that the upstream now does it in the class PaliGemmaModel and not in PaliGemmaForConditionalGeneration anymore. That's why I'm removing it from GaudiPaliGemmaForConditionalGeneration.

@schoi-habana schoi-habana mentioned this pull request Aug 22, 2025
Merged
3 tasks
@regisss regisss marked this pull request as ready for review September 3, 2025 15:14
@regisss
Copy link
Copy Markdown
Collaborator Author

regisss commented Sep 3, 2025

Merging as discussed with @astachowiczhabana. Will solve known ongoing issues in new PRs.

@regisss regisss merged commit c0fa320 into main Sep 3, 2025
5 of 9 checks passed
astachowiczhabana pushed a commit that referenced this pull request Sep 4, 2025
@regisss @IlyasMoutawwakil @astachowiczhabana
Upgrade to Transformers v4.55 (#2209)
68c459b 
Co-authored-by: IlyasMoutawwakil <ilyas.moutawwakil@gmail.com>
@astachowiczhabana
Copy link
Copy Markdown
Collaborator

astachowiczhabana commented Sep 4, 2025

LGTM

astachowiczhabana pushed a commit that referenced this pull request Sep 17, 2025
@regisss @IlyasMoutawwakil
Upgrade to Transformers v4.55 (#2209)
22a8b51 
Co-authored-by: IlyasMoutawwakil <ilyas.moutawwakil@gmail.com>
@regisss regisss mentioned this pull request Sep 25, 2025
Merged
3 tasks
gplutop7 pushed a commit to HabanaAI/optimum-habana-fork that referenced this pull request Oct 15, 2025
@HabDevops @regisss @IlyasMoutawwakil
Upgrade to Transformers v4.55 (huggingface#2209) (huggingface#648)
f428a2c 
Co-authored-by: regisss <15324346+regisss@users.noreply.github.com>
Co-authored-by: IlyasMoutawwakil <ilyas.moutawwakil@gmail.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@vivekgoe vivekgoe Awaiting requested review from vivekgoe vivekgoe is a code owner

@mandy-li mandy-li Awaiting requested review from mandy-li mandy-li is a code owner

@libinta libinta Awaiting requested review from libinta libinta is a code owner

1 more reviewer

@pbielak pbielak pbielak left review comments

Reviewers whose approvals may not affect merge requirements

Assignees

@karol-brejna-i karol-brejna-i

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

6 participants

@regisss @pbielak @HuggingFaceDocBuilderDev @astachowiczhabana @karol-brejna-i @IlyasMoutawwakil