CompRep: A Dataset For Computational Reproducibility

This curated benchmark dataset supports the evaluation and comparison of computational reproducibility tools. It comprises 18 successfully reproduced experiments from diverse scientific domains, including software engineering, AI, climate science, medical research, and economics. Each experiment package includes code, data, and comprehensive reproducibility instructions, standardized to facilitate consistent re-execution. This dataset aims to serve as a universal benchmark for researchers and practitioners working on scientific reproducibility.