text-to-image stable-diffusion

Model with best settings (VAE, scheduler, ETC) for finetuining.