The official implementation of ECCV'24 paper "To Generate or Not? Safety-Driven Unlearned Diffusion Models Are Still Easy To Generate Unsafe Images ... For Now". This work introduces one fast and effective attack method to evaluate the harmful-content generation ability of safety-driven unlearned diffusion models.
Extract API to enable unicorn. Please don't merge before we discuss. Run
uvicorn main:app --host 0.0.0.0 --port 8000under/srcto start service.