EpiCoder Meta Features is a dataset released by Microsoft in 2025, featuring a hierarchical taxonomy of meta-features and corresponding frequency statistics. The associated paper is "EpiCoder: Encompassing Diversity and Complexity in Code Generation", aiming to provide foundational support for controlled feature-conditioned code generation.

Dataset Composition

• epicoder_features.json: Defines a hierarchical taxonomy of 17 top-level code feature categories, covering dimensions such as programming languages, implementation styles, workflows, functionalities, and resource usage.
• epicoder_features_frequency.json: Provides frequency statistics for 1,795,310 feature nodes, including aggregate frequency, historical frequency, and average sub-feature frequency, to capture fine-grained code features extracted from real-world code repositories.

Citation

@inproceedings{wang2025epicoder,
  title={EpiCoder: Encompassing Diversity and Complexity in Code Generation},
  author={Yaoxiang Wang and Haoling Li and Xin Zhang and Jie Wu and Xiao Liu and Wenxiang Hu and Zhongxin Guo and Yangyu Huang and Ying Xin and Yujiu Yang and Jinsong Su and Qi Chen and Scarlett Li},
  booktitle={International Conference on Machine Learning (ICML)},
  year={2025},
  eprint={2501.04694},
  archivePrefix={arXiv},
  primaryClass={cs.CL},
  url={https://arxiv.org/abs/2501.04694},
}