Export and quantize better text-to-image model #939
Description
Our current diffusion model is quite bad and it's not worth to quantize it right now. In the future we can export better models e.g. https://huggingface.co/IDKiro/sdxs-512-dreamshaper and then return to quantizing it.