HITshan/DK-VLM-Model

DK-VLM Models

Paper

DK-VLM introduces geometry as position through Dual-Kernel Attention to improve spatial reasoning in vision-language models.

This repository provides 7B and 8B incremental checkpoints, containing LoRA adapters and geometric branch weights. Download the corresponding Qwen base model and VGGT-Omega weights separately.

See GitHub for inference and evaluation instructions.

Citation

@misc{su2026dkvlmgeometrypositionspatial,
  title={DK-VLM: Geometry as Position for Spatial Reasoning in Vision-Language Models},
  author={Han Su and Meng Tian and Tianyu Huang and Mingyu Gao and Lennox Wei and Dongxu Li and Jianhua Han and Hang Xu and Xiaohe Wu and Wangmeng Zuo},
  year={2026},
  eprint={2610.22356},
  archivePrefix={arXiv},
  primaryClass={cs.CV},
  url={https://arxiv.org/abs/2610.22356},
}