i1: Open recipe for strong text-to-image models

Arxiv CS.CV (computer vision)

Paper introduces i1, a fully open recipe for text-to-image diffusion models, including code, data, and training details. Unlike prior open-weight models, it provides a simple, reproducible baseline with limited ablations.

Text-to-imageDiffusionOpen source

See 1 more source

i1: A Simple and Fully Open-Source Recipe for Strong Text-to-Image Models2 days agoninjasaid13Discuss

Arxiv CS.CV (computer vision)·Boya ZengTianze LuoShu PuJucheng ShenTaiming LuGabriel SarchZhuang Liu

3 days ago