The dataset is available on Hugging Face in two versions:

• MedConclusion (Full 5.7M Dataset)
• MedConclusion-Compact (Fast Prototyping Dataset)

This repository contains the scripts used for generating and evaluating the conclusions:

• generate.py: Script to generate conclusions from abstract inputs using various LLMs via API.
• evaluate.py: Script to execute evaluation metrics (Rule-based scores like ROUGE/BLEU, Perplexity, and LLM-as-a-judge).

If you find this work useful, please cite:

@article{li2026medconclusion,
  title={MedConclusion: A Benchmark for Biomedical Conclusion Generation from Structured Abstracts},
  author={Li, Weiyue and Qian, Ruizhi and Li, Yi and Li, Yongce and Long, Yunfan and Cai, Jiahui and Luo, Yan and Wang, Mengyu},
  journal={arXiv preprint arXiv:2604.06505},
  year={2026}
}