Support OmniSVG

[Meatfucker]

Id love to see diffusers integration for this new and promising looking SVG model. They are already hosting the weights on huggingface.

https://github.com/OmniSVG/OmniSVG/tree/main
https://huggingface.co/OmniSVG/OmniSVG

[asomoza]

that model looks good but for what I see, the model is not a diffusion model right? SVGs are just text, so basically any LLM can do it with enough training, but this one also uses a VLLM and draw commands which make it cool, since you can see how it draws.

if is not a diffusion model, probably we can't support it.

[Meatfucker]

Oh my bad, I didnt even realize it wasnt diffusion based. My apologies.