ChemPileCap-clean

## 📘 ChemPileCap-clean This dataset is a cleaned version of [chempile-caption](https://huggingface.co/datasets/jablonkagroup/chempile-caption) Dataset. The original collection contains numerous duplicate and low-quality images, as well as a large number of erroneous or low-quality captions. We remove the duplicates and low-quality images, and further clean and re-generate highly detailed captions for the original images using gpt-4o and gemini-2.5-pro. ## 📖 Citation If you use this dataset in your research, please cite our technical report: ``` @article{fang2025uni, title={Uni-Parser Technical Report}, author={Fang, Xi and Tao, Haoyi and Yang, Shuwen and Zhong, Suyang and Lu, Haocheng and Lyu, Han and Huang, Chaozheng and Li, Xinyu and Zhang, Linfeng and Ke, Guolin}, journal={arXiv preprint arXiv:2512.15098}, year={2025} } ``` And original datasets: ``` @article{mirza2025chempile0, title = {ChemPile: A 250GB Diverse and Curated Dataset for Chemical Foundation Models}, author = {Adrian Mirza and Nawaf Alampara and Martiño Ríos-García and others}, year = {2025}, journal = {arXiv preprint arXiv:2505.12534} } ```