Preserve and Personalize: Personalized Text-to-Image Diffusion Models without Distributional Drift

Published in ICLR 2026, 2026