-
Notifications
You must be signed in to change notification settings - Fork 29.5k
chore: update model card for SigLIP #37585
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
chore: update model card for SigLIP #37585
Conversation
Hi 👋, thank you for opening this pull request! The pull request is converted to draft by default. The CI will be paused while the PR is in draft mode. When it is ready for review, please click the |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Great job, thanks so much!
docs/source/en/model_doc/siglip.md
Outdated
... torch_dtype=torch.float16, | ||
... device_map=device, | ||
... ) | ||
from transformers import SiglipModel |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Indent code here and lets show flash_attention_2
since we're already using "sdpa"
above.
# pip install -U flash-attn --no-build-isolation
<code here>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
done
Co-authored-by: Steven Liu <[email protected]>
Co-authored-by: Steven Liu <[email protected]>
Co-authored-by: Steven Liu <[email protected]>
Co-authored-by: Steven Liu <[email protected]>
Co-authored-by: Steven Liu <[email protected]>
Co-authored-by: Steven Liu <[email protected]>
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. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nice, thanks again! 🤗
- [Zero-shot image classification task guide](../tasks/zero_shot_image_classification) | ||
- Demo notebooks for SigLIP can be found [here](https://github.com/NielsRogge/Transformers-Tutorials/tree/master/SigLIP). 🌎 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@stevhliu, I'm curious, should we remove links to additional resources?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hey, great question! I think we should remove these additional resources because imo the docs should only contain official and maintained content. Everything you need to know should ideally be in the docs, and additional resources linking to extra/nice-to-know/supplemental content should be kept somewhere else. When a notebook or blog from the community gets outdated or stops working as expected, its a bit harder to maintain and can also add noise and clutter.
These additional resources are still useful and there is definitely a place for them though! I'm experimenting a bit with creating a "Transformers model resources" Space that will hold all these additional resources. Community members can open PRs to add a notebook or open fixes for it there.
Let me know if that helps answer your question! :)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@stevhliu thanks for answering, having a space indeed a cool idea! However, in my opinion, from a user perspective, it would be super nice to have some "further steps" when you land on the documentation page - so you don't need to search for additional resources and can continue exploring the model. In transformers, we are trying to keep the external API stable, so most of the resources are still relevant even though they are not maintained. Considering the case above
[Zero-shot image classification task guide](../tasks/zero_shot_image_classification)
This one is actually pointing to our internal guide
Demo notebooks for SigLIP can be found [here](https://github.com/NielsRogge/Transformers-Tutorials/tree/master/SigLIP). 🌎
This one is also still working
So, I would be more careful about removing useful materials and doing it only in case some users complain.
cc @NielsRogge as well, because you were the one who added a lot of resources
* edit siglip model card * fix syntax * Update docs/source/en/model_doc/siglip.md Co-authored-by: Steven Liu <[email protected]> * Update docs/source/en/model_doc/siglip.md Co-authored-by: Steven Liu <[email protected]> * Update docs/source/en/model_doc/siglip.md Co-authored-by: Steven Liu <[email protected]> * Update docs/source/en/model_doc/siglip.md Co-authored-by: Steven Liu <[email protected]> * Update docs/source/en/model_doc/siglip.md Co-authored-by: Steven Liu <[email protected]> * Update docs/source/en/model_doc/siglip.md Co-authored-by: Steven Liu <[email protected]> * address comments --------- Co-authored-by: Steven Liu <[email protected]>
What does this PR do?
Update the model card for SigLIP to handle #36979
Fixes #36979
Before submitting
Pull Request section?
to it if that's the case.
documentation guidelines, and
here are tips on formatting docstrings.
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.
@stevhliu