Improve HF integration

[NielsRogge]

Hi @catcathh,

Niels here from the open-source team at Hugging Face. I discovered your work through Gradio's account: https://x.com/Gradio/status/1838167906279924166 (feel free to claim authorship with your HF account). I work together with AK on improving the visibility of researchers' work on the hub.

I see you already made a demo on the 🤗 hub which is great. I've got a couple of suggestions on improving the HF integration:

• would be great to remove the checkpoints from the Space, and instead push them to a dedicated model repository, based on this guide: https://huggingface.co/docs/hub/models-uploading.

• would be great to link the demo and models to the paper: https://huggingface.co/papers/2407.02158. This can be done by simply adding the hyperlink in the README of the respective repos.

• I see there's https://huggingface.co/roubaofeipi/UltraPixel but there's no model card, would be great to link to the Github repository, adding a "Usage" example. We can then also add tags to the model repos, such as pipeline_tag: image-to-image, so that people find your models at https://huggingface.co/models?pipeline_tag=image-to-image.

Let me know if you're interested/need any help regarding this!

Cheers,

Niels ML Engineer @ HF 🤗

[catcathh]

Thanks for your suggestions! We have modified the HF space and the model repository.