CodeFeedback-Python105K Dataset Overview

Dataset Information

• Features:
  • query: string type
  • response: string type
• Splits:
  • train: contains 104,848 samples, occupying 232,791,997 bytes
• Download Size: 114,503,169 bytes
• Dataset Size: 232,791,997 bytes
• Configurations:
  • default: includes training data files data/train-*
• License: Apache 2.0
• Task Category: Question Answering
• Language: English
• Scale Category: 10K < n < 100K

Dataset Source

• This dataset is a subset extracted from the m-a-p/CodeFeedback-Filtered-Instruction dataset, which originally contains 156,526 samples.
• The original dataset includes samples from four major open‑source code instruction tuning datasets:
  • Magicoder-OSS-Instruct
  • Python code subset of ShareGPT
  • Magicoder-Evol-Instruct
  • Evol-Instruct-Code
• This subset contains only 104,848 samples written in Python.

References

@article{zheng2024opencodeinterpreter, title={Opencodeinterpreter: Integrating code generation with execution and refinement}, author={Zheng, Tianyu and Zhang, Ge and Shen, Tianhao and Liu, Xueling and Lin, Bill Yuchen and Fu, Jie and Chen, Wenhu and Yue, Xiang}, journal={arXiv preprint arXiv:2402.14658}, year={2024} }

@article{meng2024pissa, title={Pissa: Principal singular values and singular vectors adaptation of large language models}, author={Meng, Fanxu and Wang, Zhaohui and Zhang, Muhan}, journal={arXiv preprint arXiv:2404.02948}, year={4 2024} }