Training Software Engineering Agents and Verifiers with SWE-Gym

作者：Jiayi Pan, Xingyao Wang, Graham Neubig, N. Jaitly, Heng Ji, Alane Suhr, Yizhe Zhang · 发表于：International Conference on Machine Learning · 年份：2024 · DOI：10.48550/arxiv.2412.21139 · 被引用次数：288 · 研究领域：Computer Science

We present SWE-Gym, the first environment for training real-world software engineering (SWE) agents. SWE-Gym contains 2,438 real-world Python task instances, each comprising a codebase with an executable runtime environment, unit tests, and a task specified in natural language. We use SWE-Gym to train language model based SWE agents, achieving up to 19% absolute gains in resolve rate on the popular SWE-Bench Verified and Lite test sets. We also experiment with inference-time scaling through verifiers trained on agent trajectories sampled from SWE-Gym. When combined with our fine-tuned SWE agents, we achieve 32.0% and 26.0% on SWE-Bench Verified and Lite, respectively, reflecting a new state-of-the-art for open-weight SWE agents. To facilitate further research, we publicly release SWE-Gym, models, and agent trajectories.