Reproduce “Natural Language-Focused Software Engineering via Code-Documentation Equivalence” 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
Aryaz Eghbali, Zhongxin Liu, Michael Pradel
Documentary iteratively refines docstrings by asking an LLM to reconstruct function behavior from the description and file context, then using mismatches to revise it. On 238 Python functions, 127 (53.4%) passed behavioral checks, versus 81 (34.0%) for one-step generation; only 66.4% of its equivalence judgments were confirmed. In selected tests, fully correct predictions rose from 54.9% to 60.5%.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.