Scaling Concept

We use pretrained text-guided diffusion models to scale up/down concepts in image/audio.