Reproduce “Overwatch: Learning Patterns in Code Edit Sequences” 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
Yuhao Zhang, Yasharth Bajpai, Priyanshu Gupta, Ameya Ketkar, Miltiadis Allamanis, Titus Barik, Sumit Gulwani, Arjun Radhakrishna and 3 more
Overwatch learns edit-sequence patterns from developers’ IDE traces and uses recent edits to suggest related code changes. It groups recurring edits into generalized templates and matches observed sequences to predict the next edit. In Visual Studio C# traces from 12 professional developers, the selected configuration reached 78.38% precision and 36.25% relative recall; the study does not establish productivity gains or generalization beyond its setting.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.