Skip to content

Fix tiles mixing for batched input, add tie_word_embeddings to LFM2VL config#43379

Merged
zucchini-nlp merged 4 commits intohuggingface:mainfrom
Liquid4All:lfm-vl-batch-fix
Jan 28, 2026
Merged

Fix tiles mixing for batched input, add tie_word_embeddings to LFM2VL config#43379
zucchini-nlp merged 4 commits intohuggingface:mainfrom
Liquid4All:lfm-vl-batch-fix

Conversation

@ankke
Copy link
Contributor

@ankke ankke commented Jan 20, 2026

Fixes: For batched input with multiple images tiled input was mixed between batches.Now, using a util function.
Adds: tie_word_embeddings to Lfm2VlConfig. Tests for above bug + updated integration test for the newest lfm2vl model.

cc: @zucchini-nlp

Copy link
Member

@zucchini-nlp zucchini-nlp left a comment

Choose a reason for hiding this comment

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

Thanks, I will trigger slow tests and check

@zucchini-nlp
Copy link
Member

run-slow: lfm2_vl

@github-actions
Copy link
Contributor

This comment contains run-slow, running the specified jobs:

models: ["models/lfm2_vl"]
quantizations: []

@HuggingFaceDocBuilderDev

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.

@github-actions
Copy link
Contributor

CI Results

Workflow Run ⚙️

✅ No failing test specific to this PR 🎉 !

@zucchini-nlp zucchini-nlp enabled auto-merge (squash) January 27, 2026 08:34
@zucchini-nlp
Copy link
Member

I can't push to this branch apparently, can you update the docstring @ankke and add tie_word_embeddings?

@zucchini-nlp zucchini-nlp disabled auto-merge January 27, 2026 08:39
@github-actions
Copy link
Contributor

[For maintainers] Suggested jobs to run (before merge)

run-slow: lfm2_vl

@ankke ankke requested a review from zucchini-nlp January 28, 2026 01:14
Copy link
Member

@zucchini-nlp zucchini-nlp left a comment

Choose a reason for hiding this comment

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

Thanks, merging!

@zucchini-nlp
Copy link
Member

run-slow: lfm2_vl

@github-actions
Copy link
Contributor

This comment contains run-slow, running the specified jobs:

models: ["models/lfm2_vl"]
quantizations: []

@github-actions
Copy link
Contributor

CI Results

Workflow Run ⚙️

✅ No failing test specific to this PR 🎉 !

@zucchini-nlp zucchini-nlp merged commit 2a5ba8b into huggingface:main Jan 28, 2026
21 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants