Reproduce “Leveraging Generative AI for Enhancing Domain-Driven Software Design” 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
Götz-Henrik Wiegand, Filip Stepniak, Patrick Baier
Researchers fine-tuned an open-weight Large Language Model, Code Llama 7B, to automatically generate domain-specific JSON objects for Domain-Driven Design (DDD) applications. The method employed resource-efficient techniques like QLoRA and LoRA, achieving high syntactic correctness, particularly 100% for clear prompts, while operating under strict hardware and data confidentiality constraints.
Key insights
- Powerful generative AI capabilities can be made accessible for practical software engineering applications, even with stringent resource and budget constraints, through techniques like QLoRA and LoRA.
- Local fine-tuning of open-weight LLMs on proprietary datasets offers a viable solution for organizations with strict data confidentiality requirements, avoiding reliance on external cloud services.
- Automating the generation of machine-readable JSON metamodels significantly enhances the efficiency and accessibility of the Domain-Driven Design process.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.