SWE-Dev: Evaluating and Training Autonomous Feature-Driven Software Development
This addresses a prevalent real-world software engineering task for developers and AI researchers, providing a new benchmark and training platform, though it is incremental as it builds on existing LLM capabilities.
The paper tackles the underexplored problem of feature-driven development (FDD) in software engineering by introducing SWE-Dev, a large-scale dataset with 14,000 training and 500 test samples, which enables evaluation and training of autonomous coding systems; results show FDD is challenging (e.g., Claude-3.7-Sonnet achieves 22.45% Pass@3 on hard split), and fine-tuning on SWE-Dev improved a 7B model to match GPT-4o on hard tasks.
Large Language Models (LLMs) have shown strong capability in diverse software engineering tasks, e.g. code completion, bug fixing, and document generation. However, feature-driven development (FDD), a highly prevalent real-world task that involves developing new functionalities for large, existing codebases, remains underexplored. We therefore introduce SWE-Dev, the first large-scale dataset (with 14,000 training and 500 test samples) designed to evaluate and train autonomous coding systems on real-world feature development tasks. To ensure verifiable and diverse training, SWE-Dev uniquely provides all instances with a runnable environment and its developer-authored executable unit tests. This collection not only provides high-quality data for Supervised Fine-Tuning (SFT), but also enables Reinforcement Learning (RL) by delivering accurate reward signals from executable unit tests. Our extensive evaluations on SWE-Dev, covering 17 chatbot LLMs, 10 reasoning models, and 10 Multi-Agent Systems (MAS), reveal that FDD is a profoundly challenging frontier for current AI (e.g., Claude-3.7-Sonnet achieves only 22.45\% Pass@3 on the hard test split). Crucially, we demonstrate that SWE-Dev serves as an effective platform for model improvement: fine-tuning on training set enabled a 7B model comparable to GPT-4o on \textit{hard} split, underscoring the value of its high-quality training data. Code is available here \href{https://github.com/DorothyDUUU/SWE-Dev}{https://github.com/DorothyDUUU/SWE-Dev}.