AI RESEARCH

SHIFT: Steering Hidden Intermediates in Flow Transformers

arXiv CS.CV

ArXi:2604.09213v1 Announce Type: new Diffusion models have become leading approaches for high-fidelity image generation. Recent DiT-based diffusion models, in particular, achieve strong prompt adherence while producing high-quality samples. We propose SHIFT, a simple but effective and lightweight framework for concept removal in DiT diffusion models via targeted manipulation of intermediate activations at inference time, inspired by activation steering in large language models.