The root cause is that the text encoders (e.g. CLIPText) used to power image generation models are too good and can handle semantic synonyms, even oblique synon

The root cause is that the text encoders (e.g. CLIPText) used to power image generation models are too good and can handle semantic synonyms, even oblique synonyms such as "Mickey Mouse" and "a famous anthropomorphic mouse mascot". The repo mentioned focuses on NSFW content, but I've had an idea that can theoretically identify arbitrary cases that could be undesirable by image generation services: get a large dataset of CLIPText encoded texts, and train a LLM to take in said embeddings and output synonyms, or maybe something with contrastive loss like CLIP itself.