EZBugs4Py: A benchmark of simple, easily reproducible Python bugs
Gábor Antal, Norbert Vándor, Rudolf Ferenć · Zenodo (CERN European Organization for Nuclear Research) · 2023
This is the appendix for paper entitled "EZBugs4Py: A benchmark of simple, easily reproducible Python bugs" submitted to MSR 2024. The dataset itself is available on GitHub. This appendix contains: The results of GPT in the following tasks: automated program repair for to so-called "buggy" versions of the programs, the "failing" versions of the program, and the code synthesis based on the descriptions of the tasks. The exact prompts we used in the paper. The runner scripts to query GPT-4. The categorization of the bugs.