With transformer based models becoming defacto for the diffusion community, I think it makes sense to provide support for saving and loading the quantized models in diffusers through optimum.quanto...