Reproduce “An Empirical Study on the Potential of LLMs in Automated Software Refactoring” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Bo Liu, Yanjie Jiang, Yuxia Zhang, Nan Niu, Guangjie Li, Hui Liu
The authors evaluate LLMs for finding and recommending Java refactorings and introduce RefactoringMirror, which detects edits in an LLM’s output and reapplies recognized operations through IntelliJ. On 180 known opportunities, specific prompts raised GPT-4’s identification success rate to 86.7%. RefactoringMirror avoided observed bugs in 22 buggy outputs, but results are limited to two models and Java.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.