1 paper · 1 filter
Zixiang Li, Yue Song, Renshuai Tao +3
Although diffusion models have achieved remarkable success in the field of image generation, their latent space remains under-explored. Current methods for identifying semantics wi…