PA-LLaVA-plus

This is the first-stage weights trained on the 400w pathological image-text dataset using the PA-LLaVA model structure. The link is https://huggingface.co/OpenFace-CQUPT/PA-LLaVA-plus.

400w Dataset

The 400w pathology dataset is derived from the publicly available "Accessible Dataset (18M samples)" from MedTrinity-25M(UCSC-VLAA/MedTrinity-25M · Datasets at Hugging Face). This is a dataset spanning multiple medical fields. By analyzing the linguistic structure of the text in this dataset, we extracted a 400w dataset specific to the pathology domain.

Citation

@INPROCEEDINGS{10821785,
  author={Dai, Dawei and Zhang, Yuanhui and Xu, Long and Yang, Qianlan and Shen, Xiaojing and Xia, Shuyin and Wang, Guoyin},
  booktitle={2024 IEEE International Conference on Bioinformatics and Biomedicine (BIBM)}, 
  title={PA-LLaVA: A Large Language-Vision Assistant for Human Pathology Image Understanding}, 
  year={2024},
  volume={},
  number={},
  pages={3138-3143},
  keywords={Connectors;Pathology;Visualization;Codes;Computational modeling;Biological system modeling;Data models;Cleaning;Bioinformatics;Biomedical imaging;Pathology Image Understanding;VQA;LLaVA},
  doi={10.1109/BIBM62325.2024.10821785}}

@article{dai2025pathologyvlm,
  title={Pathologyvlm: a large vision-language model for pathology image understanding},
  author={Dai, Dawei and Zhang, Yuanhui and Yang, Qianlan and Xu, Long and Shen, Xiaojing and Xia, Shuyin and Wang, Guoyin},
  journal={Artificial Intelligence Review},
  volume={58},
  number={6},
  pages={1--19},
  year={2025},
  publisher={Springer}
}

---

license: cc