Publications
Preprints
Zoom in, Click out: Unlocking and Evaluating the Potential of Zooming for GUI Grounding
arXiv preprint (under review), 2025
A training-free method that leverages zooming as a powerful prior for GUI grounding, achieving state-of-the-art results across multiple benchmarks.
Wow, wo, val! A Comprehensive Embodied World Model Evaluation Turing Test
arXiv preprint (under review), 2026
The first comprehensive benchmark for embodied world models, examining five core abilities with 22 evaluation metrics and strong correlation with human preference.
WoW: Towards a World omniscient World model Through Embodied Interaction
arXiv preprint (under review), 2025
A 14-billion parameter world model trained on 2 million robot trajectories, demonstrating that physical intuition emerges from large-scale embodied interaction.