In this replication package, we describe how to replicate the results of our ICSE’23 paper, [Large Language Models are Few-shot Testers: Exploring LLM-based General Bug Reproduction]. Our replication package allows for an accurate reconstruction of the figures and tables presented within our paper, as well as the source code for the tool that we built to generate bug reproducing tests. Meanwhile, our tool used closed API access to a large language model, Codex, which necessitates corresponding access tokens to replicate the test generation aspect of our work. In lieu of providing our API key, we provide 50 generated tests for each bug we replicate, which allows downstream replication of our results.

We claim the Artifacts Available and Artifacts Reusable badge. A reviewer is assumed to be familiar with Docker, Python, and Jupyter Notebooks for our package.