Reproduce “Experimental Settings in LLM-Based Program Repair: A Study of Inputs, Tool Access, Feedback, and Validation” 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
Xushu Dai, Yicheng Cai, Nanqing Luo, Pei-Yu Tseng
The authors introduce a machine-readable record of automated program repair evaluation conditions, including inputs, fault localization, permitted tools, feedback, and final tests. Their review of 10 settings from seven studies finds different tasks within the same benchmark family; six settings return execution or tool output. Scores compare complete systems under reported conditions, not models or methods when those conditions differ.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.