LookAgain-8B
Description
This is the official model weights of LookAgain: Closed-Loop GUI Grounding with Visually Grounded Reflection. In this work, we introduce LookAgain, a novel closed-loop grounding paradigm that enables MLLM-based GUI agents to iteratively predict, inspect, and refine coordinates through visually grounded reflection. By combining reflective trajectory SFT with GRPO optimization, LookAgain achieves robust GUI grounding performance on various challenging scenarios.
How to Run?
Please refer to the instructions in the Githhub repository.
Citation
If you find this work useful for your research, please kindly cite our paper:
@article{zhang2026lookagain,
title={LookAgain: Closed-Loop GUI Grounding with Visually Grounded Reflection},
author={Zhang, Renshan and Meng, Haoyang and He, Yixiao and Shao, Rui and Liu, April Hua and Nie, Liqiang},
journal={arXiv preprint arXiv:2608.09723},
year={2026}
}