add qwen2.5vl #35569

ArthurZucker merged 46 commits into huggingface:main from ShuaiBai623:add_qwen2.5vl

ShuaiBai623177 days ago🎉 16

What does this PR do?

Fixes # (issue)

Before submitting

This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
Did you read the contributor guideline,
Pull Request section?
Was this discussed/approved via a Github issue or the forum? Please add a link
to it if that's the case.
Did you make sure to update the documentation with your changes? Here are the
documentation guidelines, and
here are tips on formatting docstrings.
Did you write any new necessary tests?

Who can review?

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.

add qwen2.5vl

e693210a

ShuaiBai623 requested a review from

ArthurZucker 177 days ago

ShuaiBai623 requested a review from

qubvel 177 days ago

ShuaiBai623 requested a review from

molbap 177 days ago

ShuaiBai623 requested a review from

yonigozlan 177 days ago

ShuaiBai623 requested a review from

stevhliu 177 days ago

ShuaiBai623 requested a review from

Rocketknight1 177 days ago

fix

8a713a90

pass check table

be1f811c

minostauros requested changes on 2025-01-09

minostauros177 days ago

Brought changes from #35466
Thanks for the new model!

Conversation is marked as resolved

Show resolved

Conversation is marked as resolved

Show resolved

Conversation is marked as resolved

Show resolved

qubvel added New model

qubvel added Multimodal

add modular file

a666eb02

fix style

a96b32d6

Update src/transformers/models/qwen2_5_vl/modeling_qwen2_5_vl.py

8191ce9b

Update src/transformers/models/qwen2_5_vl/modeling_qwen2_5_vl.py

2b70136f

Update src/transformers/models/qwen2_5_vl/modeling_qwen2_5_vl.py

c23055b7

padd copy check

b1e2f688

use modular

9d0e553e

ShuaiBai623 requested a review from

zucchini-nlp 169 days ago

fix

23e6635c

fix

aa0dc95c

fix

05a1ab15

update flashatt2&sdpa support_list

00b05f2d

stevhliu commented on 2025-01-17

stevhliu168 days ago

Thanks :)

Update docs/source/en/_toctree.yml

47eb0694

Update docs/source/en/model_doc/qwen2_5_vl.md

35b9dbb2

Update docs/source/en/model_doc/qwen2_5_vl.md

67a10dae

Update docs/source/en/model_doc/qwen2_5_vl.md

e37ce587

Update docs/source/en/model_doc/qwen2_5_vl.md

fcbb355a

Update src/transformers/models/qwen2_5_vl/modular_qwen2_5_vl.py

7f3c66e7

update config

f5a4c2b5

Merge branch 'main' into add_qwen2.5vl

00f9832f

update

f57827dd

fix hf path

b21bd62d

rename Qwen2_5_VLVideosKwargs

18ea4bef

gewenbin0992 force pushed to 18ea4bef 166 days ago

fix

faae9f48

fix

cd3fd21c

ArthurZucker commented on 2025-01-20

ArthurZucker166 days ago

Super super nice! 🤗
Great use of modular, thanks for the various integration tests as well!

update

f8c465cb

excuted modular

84247a47

rollback init

0cd1f621

fix

446383fe

formated

f87160db

simpler init

4012f4aa

fix

d671be8c

fix

ad46b878

fix

8390d99d

fix

16913ab0

fix

72de3020

update docs

91a00e32

minostauros commented on 2025-01-22

minostauros164 days ago

Again, brought changes from #35466

fix

ca2e16cd

fix

64db9c7d

Merge branch 'huggingface:main' into add_qwen2.5vl

bb0c4596

Merge branch 'main' into add_qwen2.5vl

df50ed56

update Qwen2VLRotaryEmbedding for yarn

c84216d7

ArthurZucker approved these changes on 2025-01-23

ArthurZucker163 days ago😄 1

Thanks only a few comments that are not adressed, we like to split long sequence of code in more line for debugging purpose and readability!

Thanks for your PR 🤗

Merge branch 'main' into add_qwen2.5vl

87882dbe

fix

a0b3a945

ArthurZucker merged f3f6c865 into main 163 days ago

Reviewers

ArthurZucker

minostauros

stevhliu

gewenbin0992

qubvel

molbap

yonigozlan

Rocketknight1

zucchini-nlp

Assignees

No one assigned

Labels

New model Multimodal

Milestone

No milestone

	1502
	1503		def get_rope_index(
	1504		self,
	1505		input_ids: torch.LongTensor,

	1661		if attention_mask is not None:
	1662		position_ids = attention_mask.long().cumsum(-1) - 1
	1663		position_ids.masked_fill_(attention_mask == 0, 1)
	1664		position_ids = position_ids.unsqueeze(0).expand(3, -1, -1).to(input_ids.device)

	1788		attention_mask = attention_mask.to(inputs_embeds.device)
	1789
	1790		# if we get 4D attention mask we cannot calculate rope deltas anymore. TODO @raushan fixme
	1791		if position_ids is None and input_ids is not None and (attention_mask is None or attention_mask.ndim == 2):

	input_ids: torch.LongTensor,
	input_ids: Optional[torch.LongTensor] = None,

	if position_ids is None and input_ids is not None and (attention_mask is None or attention_mask.ndim == 2):
	if position_ids is None and (attention_mask is None or attention_mask.ndim == 2):

transformers add qwen2.5vl #35569 Merged

add qwen2.5vl #35569

What does this PR do?

Before submitting

Who can review?

transformers
add qwen2.5vl
#35569

Merged