KLOD: Locality-Preserving Knowledge Editing via Non-Target Distribution Preservation
KLOD is proposed, a bounded and distribution-preserving objective for fine-tuning-based knowledge editing that separates the intended target update from distributions that should remain stable and substantially mitigates locality degradation while maintaining high edit reliability.
Hojun Jeong, Gyunyeop Kim, Sangwoo Kang
· 0 citations