Paper summary
Paper summary: CLEAR: Character Unlearning in Textual and Visual Modalities
CLEAR is an open benchmark for removing information about specific people from multimodal models, with 200 fictitious individuals, 3,700 images and linked Q&A, used to compare 11 unlearning methods.
- The people are fictitious, so "forgetting" can be measured cleanly without real personal data.
- Unlearning both modalities together beat unlearning text or images alone.