Add Lumina-Image-2.0 Image Generative Model

[SuperPat45]

Add the new Lumina-Image-2.0 Image Generative Model.
https://huggingface.co/Alpha-VLLM/Lumina-Image-2.0
It generates very good quality images and is Apache licensed.

Source code:
https://github.com/Alpha-VLLM/Lumina-Image-2.0

[mudler]

still not available in diffusers:

ImportError: cannot import name 'Lumina2Text2ImgPipeline' from 'diffusers' (/home/runner/work/LocalAI/LocalAI/backend/python/diffusers/venv/lib/python3.10/site-packages/diffusers/__init__.py)

But I've prepared #4806 - once diffusers will tag a new release we will support it!

[SuperPat45]

Diffusers 0.33.0 released: https://github.com/huggingface/diffusers/releases/tag/v0.33.0

This version also adds Sana-Sprint and OmniGen pipelines

[SuperPat45]

"Lumina" model does not return anything in the gallery

[Hello-World-Traveler]

914 models from 1 repositories, is this correct? other then missing this model.

[SuperPat45]

https://huggingface.co/Alpha-VLLM/Lumina-Image-2.0/tree/main/transformer
They also have a 2b parameters:
https://huggingface.co/Alpha-VLLM/Lumina-Next-SFT-diffusers/tree/main/transformer

[Hello-World-Traveler]

Has this been held back?

[SuperPat45]

I couldn't get it to work, could someone share their yaml description file with the links to the files that were used?

[Hello-World-Traveler]

The model was never added, what was added was the backend Lumina2Text2ImgPipeline

in the model card use pipeline_type: Lumina2Text2ImgPipeline