EAGLE: Expert‑Guided Self‑Enhancement for Preference Alignment in Pathology Large Vision‑Language Models (ACL2025)

EAGLE is a three-stage alignment framework designed to enhance Large Vision-Language Models (LVLMs) in pathology by leveraging expert-guided self-enhancement and preference data.

---

💡 Key Contributions

• Introduces a scalable, expert-guided self-enhancement pipeline.
• Builds pathology-specific preference pairs with minimal human cost.
• Improves faithfulness, factual accuracy, and localization ability in pathology VQA tasks.

---

More information

For code, weights, etc, please see here.

📎 Citation

@inproceedings{ding2025eagle,
  title={EAGLE: Expert-Guided Self-Enhancement for Preference Alignment in Pathology Large Vision-Language Model},
  author={Ding, Meidan and Zhang, Jipeng and Wang, Wenxuan and Zhong, Haiqin and Wang, Xiaoqin and Lyu, Xinheng and Chen, Wenting and Shen, Linlin},
  booktitle={Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  pages={14603--14619},
  year={2025}
}