Memory-efficient Diffusion Transformers with Quanto and Diffusers