Reproduce “Coeditor: Leveraging Contextual Changes for Multi-round Code Auto-editing” 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
Jiayi Wei, Greg Durrett, Isil Dillig
Coeditor helps developers edit existing code by conditioning suggestions on recent repository changes and relevant code context. It represents edits as line-based additions and deletions, and uses static analysis to assemble related code context. On PYCOMMITS-ONELINE, its 220-million-parameter model achieves 60.4% enhanced exact match, versus 39.5% for text-davinci-003. This result concerns predicting one changed line, not repository-wide editing.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.