How to create Inference API for custom model?

Hi, I have uploaded my custom model with novel architecture for HF hub. It provides image classification but by default no Inference API was provided.
Can I somehow create this Inference mode manually?

is your model public? post your model here if public - it may help someone find the issue for you quicker.

Yes. My model is public.


The inference API assumes a corresponding pipeline for it is defined. So we will first have to merge CLIPForImageClassification has been added by Andron00e · Pull Request #27805 · huggingface/transformers · GitHub