i1: Fully open recipe for text-to-image models

Arxiv CS.CV (computer vision)

arXiv paper presents i1, a fully open-source recipe for text-to-image diffusion models, with detailed ablations on architecture and data. Code and model weights are released to enable reproducible progress.

Diffusion modelsText-to-image

See 1 more source

i1: A Simple and Fully Open-Source Recipe for Strong Text-to-Image Models8 days agoninjasaid13Discuss

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

Jun 11, 4:00 AM