1 paper
Daiheng Gao, Shilin Lu, Shaw Walters +8
Removing unwanted concepts from large-scale text-to-image (T2I) diffusion models while maintaining their overall generative quality remains an open challenge. This difficulty is es…