DanceOPD: On-Policy Generative Field Distillation
A novel on-policy generative field distillation framework called DanceOPD is proposed to unify text-to-image generation, local editing, and global editing capabilities in flow-matc…
Hugging Face · Daily Papers
·Wei Zhou, Xiongwei Zhu
·
·▲ 64 upvotes
Este artigo está em destaque na seleção diária de papers do Hugging Face, curada pela comunidade de pesquisa em IA.
Autores: Wei Zhou, Xiongwei Zhu, Zelin Xu, Bo Dong, Lixue Gong, Yongyuan Liang
- 64 upvotes da comunidade
- Temas: generative field distillation, flow-matching models, on-policy, velocity MSE objective, velocity field, classifier-free guidance
Resumo
Resumo original (em inglês), extraído do paper:
A novel on-policy generative field distillation framework called DanceOPD is proposed to unify text-to-image generation, local editing, and global editing capabilities in flow-matching models through capability-specific routing and velocity-based training.
