SanghyukChun

/

LongProLIP-ViT-B-16-SHD128M

like 0

@nielsr
Thanks for your PR. When I checked the image-to-text pipeline, it does not fit my model (it is more like CLIP or SigLIP, not a captioning model). Instead, I will add pipeline Zero-Shot Image Classification for all the PRs you suggested.

Best,
Sanghyuk

I just found that transformer library is also not a proper tag (I built my own library based on open_clip, and my library cannot be directly used with transformer or open_clip, but it has its own library prolip). I remove the transformer library tags and changed image-to-text to Zero-Shot Image Classification.

Thanks for your help again.

Thanks! In that case you can add library_name: prolip. If you're interested, you can also submit a PR like this one: https://github.com/huggingface/huggingface.js/pull/1264.

This ensures that the hub recognizes "prolip" as a library, and it will provide an appropriate code snippet (which you can customize) if you click on "Use this model" at the top right of a model.

Thanks. I added library_name: prolip to the models in the ProLIP collection. I will consider submitting a PR as well.

Thanks!