SWE-Synth: Synthesizing Verifiable Bug-Fix Data to Enable Large Language Models in Resolving Real-World Bugs
This addresses a bottleneck in software engineering automation for developers by providing scalable synthetic data, though it is incremental as it builds on existing agent-based approaches.
The paper tackles the lack of high-quality training datasets for large language models in automated program repair by introducing SWE-Synth, a framework that synthesizes realistic and verifiable bug-fix data, resulting in models trained on this data outperforming those on real-world datasets by 2.3% on SWE-Bench Lite.
Large language models (LLMs) are transforming automated program repair (APR) through agent-based approaches that localize bugs, generate patches, and verify fixes. However, the lack of high-quality, scalable training datasets, especially those with verifiable outputs and intermediate reasoning traces-limits progress, particularly for open-source models. In this work, we present SWE-Synth, a framework for synthesizing realistic, verifiable, and process-aware bug-fix datasets at the repository level. SWE-Synth leverages LLM agents to simulate debugging workflows, producing not only bug-fix pairs but also test cases and structured repair trajectories. Compared to manually curated datasets, our method scales with minimal human effort while preserving contextual richness and correctness. Experiments show that models trained on SWE-Synth outperform those trained on real-world datasets by 2.3% on SWE-Bench Lite. Our results highlight the potential of synthetic, agent-generated data to advance the state of the art in APR and software engineering automation.