← back to paper
arxiv: 2606.22568 · 2 revisions
SeFi-Image: A Text-to-Image Foundation Model with Semantic-First Diffusion