Free Lunch for Unified Multimodal Models: Enhancing Generation via Reflective Rectification with Inherent Understanding

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Jiang, Yibo, Wu, Tao, Jiang, Rui, Lu, Yehao, Cai, Chaoxiang, Qin, Zequn, Li, Xi
Format:	Preprint
Published:	2026
Subjects:	Computer Vision and Pattern Recognition Artificial Intelligence
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Internet

https://arxiv.org/abs/2604.13540