CReSt

# Dataset for CReSt This repository contains the dataset used in the paper "CReSt: A Comprehensive Benchmark for Retrieval-Augmented Generation with Complex Reasoning over Structured Documents". ## 📂 Dataset Overview This dataset consists of the following two subsets: - refusal: The subset that contains only refusal cases - non_refusal: The subset that contains only non-refusal cases ## 🚀 How to Use You can access detailed usage instructions in the [CReSt GitHub repository](https://github.com/UpstageAI/CReSt). Simply clone the repository and follow the steps provided in the README there: ```bash git clone https://github.com/UpstageAI/CReSt ``` # 📜 License This dataset is distributed under the CC-by-NC 4.0. # 📝 Citation If you use this dataset in your research, please cite: ```css @inproceedings{khang2025crest, title={CReSt: A Comprehensive Benchmark for Retrieval-Augmented Generation with Complex Reasoning over Structured Documents}, author={Khang, Minsoo and Park, Sangjun and Hong, Teakgyu and Jung, Dawoon}, booktitle={TBD}, pages={TBD}, year={2025} } ```