Reproduce “LLM4MTLs: Automated Generation and Empirical Evaluation of Model Transformation Languages” 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

Bowen Jiang, Nathan Hagel, Haowei Cheng, Benedikt Jutz, Arne Lange, Weixing Zhang, Rahul Sharma, Ralf Reussner and 1 more

LLM4MTLs evaluates prompt strategies for generating transformations in four model transformation languages. It builds prompts from task descriptions and metamodel context, optionally adding three examples or grammar. Few-shot prompting improved syntax across four languages, but semantic gains varied. For QVTo, Claude Sonnet 4.5’s Pass@1 rose from 0.00 to 0.70; ATL remained at 0.67. Scores cover parsable first candidates passing tests.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.