Responsible Diffusion Models via Constraining Text Embeddings within Safe Regions

Open in new window