Reproduce “Generating High-Quality Datasets for Code Editing via Open-Source Language Models” 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
Zekai Zhang, Mingwei Liu, Zhenxi Chen, Linxi Liang, Yuxuan Chen, Guangsheng Ou, Yanlin Wang, Dan Li and 2 more
OpenCodeEdit creates synthetic instruction-guided code-editing examples and filters them for fine-tuning language models. Its pipeline pairs code with concise and detailed requests and revised code, then removes unchanged, complex, and overrepresented examples. On CanItEdit, OpenCodeEdit-Qwen3-8B reaches 54.10% overall pass@1, versus 33.31% for Qwen3-8B-Instruct; this result is limited to the tested Python benchmark.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.