构建20万条多语言编程轨迹数据集，助力开源智能体实现长周期推理。
Open-SWE-Traces: Advancing Dual-Mode Multilingual Distillation for Software Engineering Agents
- 融合显式思维与非思维轨迹，用混合推理生成高质量编程数据。
- 在SWE-bench上达成61.7%的解决率，支持多语言和复杂任务。
- 适合开发高效开源编程智能体的研究者与开发者使用。
自主软件工程的发展受限于多样且大规模的轨迹数据匮乏。本文提出 ewdataset，一个包含207,489条跨九种编程语言（Python、Go、TS、JS、Rust、Java、PHP、C、C++）的智能体轨迹数据集，数据源自20,000个真实代码提交，通过OpenHands与SWE-agent工具链获取。采用混合推理合成策略：Minimax-M2.5生成带显式“思考”过程的轨迹，Qwen3.5-122B生成高质量“非思考”轨迹。数据经许可筛选（MIT、Apache、BSD），来自SWE-rebench-V2。该数据集可支持长周期推理模型训练。通过微调Qwen3-30B-A3B系列（Thinking、Instruct、Coder）验证其有效性，最佳模型在SWE-bench Verified上达到61.7%解决率，SWE-bench Multilingual为57.1%，SWE-bench Pro为36.8%。结果表明，Open-SWE-Traces是将人类级软件工程能力蒸馏至高效开源智能体的重要资源。
原文摘要 · Abstract (English)
The path toward autonomous software engineering is currently bottlenecked by a severe deficit of diverse, large-scale trajectory data. We address this by introducing \ourdataset, an expansive dataset of 207,489 agentic trajectories spanning nine programming languages (Python, Go, TS, JS, Rust, Java, PHP, C, C++). Sourced from 20,000 real-world PRs via OpenHands and SWE-agent harnesses, the dataset utilizes a hybrid-reasoning synthesis: Minimax-M2.5 generates trajectories with explicit "thinking" processes, while Qwen3.5-122B provides high-quality "non-thinking" traces. Filtered for permissive licenses (MIT, Apache, BSD) from SWE-rebench-V2, this data facilitates the training of models capable of long-horizon reasoning. We validate the dataset by fine-tuning the Qwen3-30B-A3B series (Thinking, Instruct, and Coder). The best performing model achieves resolve rates of 61.7% on SWE-bench Verified, 57.1% on SWE-bench Multilingual, and 36.8% on SWE-bench Pro. These results establish Open-SWE-Traces as a premier resource for distilling human-level software engineering capabilities into efficient, open-source agentic LLMs.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。