Safe Few-Step Generation via Velocity Editing
VESFlow is a training-free safety method for flow matching-based text-to-image generation that edits velocity fields to ensure safe output while maintaining prompt integrity.
Hugging Face · Daily Papers
·Yujin Choi, Jaehong Yoon
·
·▲ 4 upvotes
Este artigo está em destaque na seleção diária de papers do Hugging Face, curada pela comunidade de pesquisa em IA.
Autores: Yujin Choi, Jaehong Yoon
- 4 upvotes da comunidade
- Temas: flow matching, text-to-image generation, safety, concept removal, marginal velocity, velocity field
Resumo
Resumo original (em inglês), extraído do paper:
VESFlow is a training-free safety method for flow matching-based text-to-image generation that edits velocity fields to ensure safe output while maintaining prompt integrity.
// relacionados
Leia também
Blog
ByteDance's Seedance 2.5 breaks the 30-second barrier for AI video generation
Blog
Hierarchical Pooling for Sheaf Neural Networks
Blog
JPPD: Joint Prediction_Planning Diffusion with Differentiable Safety Guidance for Dynamic Obstacle Avoidance in Intelligent Transportation Systems
Blog