Reproduce “Knowledge-Graph-Driven Data Synthesis for Low-Resource Software Development: A HarmonyOS Case Study” 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
Mingwei Liu, Zheng Pei, Yanlin Wang, Zihao Wang, Zikang Li, Enci Lin, Xin Peng, Zibin Zheng
The authors turn API documentation into single- and multi-API questions and code answers, using an API graph and uncertainty-guided Monte Carlo tree search to select combinations, then fine-tune a model on examples. On a HarmonyOS benchmark, Qwen2.5-Coder-7B reaches 25.00% pass@1, versus 10.19% with OSS-Instruct data at equal scale; this result is limited to that benchmark and setup.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.