Training Software Engineering Agents and Verifiers with SWE-Gym
This addresses the need for better automated software engineering tools for developers, but it is incremental as it builds on existing language model methods.
They tackled the problem of training software engineering agents by introducing SWE-Gym, a new environment with real-world Python tasks, and achieved up to 19% absolute gains in resolve rate and a new state-of-the-art of 32.0% on SWE-Bench Verified.
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.