AdaVD removes target concepts from diffusion models by soft-projecting value vectors away from the target token direction, with a sigmoid threshold that preserves unrelated prompts.
https://huggingface.co/Lykon/ DreamShaper, 2023
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Precise, Fast, and Low-cost Concept Erasure in Value Space: Orthogonal Complement Matters
AdaVD removes target concepts from diffusion models by soft-projecting value vectors away from the target token direction, with a sigmoid threshold that preserves unrelated prompts.