Automatic Program Repair with OpenAI's Codex: Evaluating QuixBugs

Julian Aron Prenner; Romain Robbes

Automatic Program Repair with OpenAI's Codex: Evaluating QuixBugs

Software Engineering 2021-11-09 v1

Authors: Julian Aron Prenner , Romain Robbes

Abstract

OpenAI's Codex, a GPT-3 like model trained on a large code corpus, has made headlines in and outside of academia. Given a short user-provided description, it is capable of synthesizing code snippets that are syntactically and semantically valid in most cases. In this work, we want to investigate whether Codex is able to localize and fix bugs, a task of central interest in the field of automated program repair. Our initial evaluation uses the multi-language QuixBugs benchmark (40 bugs in both Python and Java). We find that, despite not being trained for APR, Codex is surprisingly effective, and competitive with recent state of the art techniques. Our results also show that Codex is slightly more successful at repairing Python than Java.

Keywords

bug localization software refactoring code generation

Cite

@article{arxiv.2111.03922,
  title  = {Automatic Program Repair with OpenAI's Codex: Evaluating QuixBugs},
  author = {Julian Aron Prenner and Romain Robbes},
  journal= {arXiv preprint arXiv:2111.03922},
  year   = {2021}
}

Related papers

View all related →