AweAI-Team/Scale-SWE-Distilled

# Immersion in the GitHub Universe: Scaling Coding Agents to Mastery

<div align="center">

[](https://arxiv.org/abs/2602.09892)
[](https://github.com/AweAI-Team/ScaleSWE)
[](https://huggingface.co/collections/AweAI-Team/scale-swe)
[](https://huggingface.co/Awe-AI/Scale-SWE)
[](LICENSE)
<br>

</div>

## 🔥 Highlights

- Source from 6M+ pull requests and 23000+ repositories.
- Cover 5200 Repositories.
- 100k high-quality instances.
- 71k trajectories from DeepSeek v3.2 with 3.5B token.
- Strong performance: 64% in SWE-bench-Verified trained from Qwen3-30A3B-Instruct.

## 📣 News
- **2026-02-26** 🚀 We released a portion of our data on [Hugging Face](https://huggingface.co/collections/AweAI-Team/scale-swe). This release includes **20,000 SWE task instances**—currently the largest **Real Executable** open-source SWE dataset available—alongside **71k distillation trajectories(3.5B)** from DeepSeek v3.2. **Much more data** will be released in the future.
- **2026-02-10** 📝 Our paper [**"Immersion in the GitHub Universe: Scaling Coding Agents to Mastery"**](https://arxiv.org/abs/2602.09892) is now available on arXiv.

## 📊 Data Format

| Field | Description |
| :--- | :--- |
| **`instance_id`** | A unique identifier formatted as `{user}_{repo}_pr{id}`. |
| **`user`** | The owner of the GitHub repository. |
| **`repo`** | The name of the GitHub repository. |
| **`language`** | The programming language of the codebase (currently Python). |
| **`workdir`** | The working directory path within the environment. |
| **`image_url`** | The URL of the pre-built Docker image for the task. |
| **`patch`** | The ground-truth patch (**Golden Patch**) from the corresponding pull request. |
| **`pr_commit`** | The commit hash of the pull request. |
| **`parent_commit`** | The commit hash of the parent commit (base state). |
| **`problem_statement`** | The issue description conveying the bug, provided to the model as input. |
| **`f2p_patch`** | The developer-written test patch containing tests that fail before the fix (if available). |
| **`f2p_script`** | The synthetic reproduction script generated by our unit-test creator agent. |
| **`FAIL_TO_PASS`** | Unit tests that fail on the buggy version but pass after the fix. |
| **`PASS_TO_PASS`** | Unit tests that pass in both versions (regression tests). |
| **`github_url`** | The URL of the original GitHub repository. |
| **`pre_commands`** | These commands must be executed immediately upon entering the container to check out the correct commit. |

## 🤖 Results
We fine-tuned Qwen-30B-A3B-Instruct on our synthesized trajectories.

## 📖 Citation

If you find this project useful for your research, please consider citing our paper:
```
@misc{zhao2026immersiongithubuniversescaling,
title={Immersion in the GitHub Universe: Scaling Coding Agents to Mastery},
author={Jiale Zhao and Guoxin Chen and Fanzhe Meng and Minghao Li and Jie Chen and Hui Xu and Yongshuai Sun and Xin Zhao and Ruihua Song and Yuan Zhang and Peng Wang and Cheng Chen and Jirong Wen and Kai Jia},
year={2026},
eprint={2602.09892},
archivePrefix={arXiv},
primaryClass={cs.SE},
url={https://arxiv.org/abs/2602.09892},
}
```

## 📄 License

This project is licensed under the CC BY 4.0 License - see the [LICENSE](LICENSE) file for details.

# 深入GitHub全域：规模化编码智能体迈向精通

<div align="center">

[](https://arxiv.org/abs/2602.09892)
[](https://github.com/AweAI-Team/ScaleSWE)
[](https://huggingface.co/collections/AweAI-Team/scale-swe)
[](https://huggingface.co/Awe-AI/Scale-SWE)
[](LICENSE)
<br>

</div>

## 🔥 核心亮点

- 数据集源自600万+拉取请求（Pull Request，PR）与2.3万余个代码仓库。
- 覆盖5200个代码仓库。
- 包含10万个高质量样本实例。
- 拥有7.1万条由DeepSeek v3.2生成的轨迹数据，总Token数达35亿。
- 性能表现优异：基于Qwen3-30A3B-Instruct微调的模型在SWE-bench-Verified基准上取得64%的准确率。

## 📣 最新动态
- **2026-02-26** 🚀 我们已在Hugging Face平台发布部分数据集。本次发布包含**2万个软件工程师任务实例**——这是目前已开源的规模最大的**可真实执行**的SWE（Software Engineering）数据集，同时附带7.1万条由DeepSeek v3.2生成的蒸馏轨迹数据（总Token数达35亿）。未来我们还将发布更多数据。
- **2026-02-10** 📝 我们的论文《**深入GitHub全域：规模化编码智能体迈向精通**》现已在arXiv平台上线。

## 📊 数据格式

| 字段 | 描述 |
| :--- | :--- |
| **`instance_id`** | 格式为`{user}_{repo}_pr{id}`的唯一标识符。 |
| **`user`** | GitHub代码仓库的所有者账号。 |
| **`repo`** | GitHub代码仓库的名称。 |
| **`language`** | 代码库所用的编程语言（当前仅支持Python）。 |
| **`workdir`** | 运行环境内的工作目录路径。 |
| **`image_url`** | 用于该任务的预构建Docker镜像的URL。 |
| **`patch`** | 对应拉取请求中的真实修复补丁（**黄金补丁（Golden Patch）**）。 |
| **`pr_commit`** | 拉取请求对应的提交哈希值。 |
| **`parent_commit`** | 父提交的哈希值（即任务的基准状态）。 |
| **`problem_statement`** | 用于向模型输入的、描述程序缺陷的问题说明。 |
| **`f2p_patch`** | 开发者编写的测试补丁，包含修复前会失败的测试用例（若有提供）。 |
| **`f2p_script`** | 由我们的单元测试生成智能体生成的人工可复现脚本。 |
| **`FAIL_TO_PASS`** | 在存在缺陷的代码版本中失败，但在修复后成功通过的单元测试。 |
| **`PASS_TO_PASS`** | 在两个版本中均能正常通过的单元测试（即回归测试用例）。 |
| **`github_url`** | 原始GitHub代码仓库的URL。 |
| **`pre_commands`** | 进入容器后需立即执行的命令，用于切换至正确的提交版本。 |

## 🤖 实验结果
我们基于自研轨迹数据对Qwen-30B-A3B-Instruct模型进行了微调。

## 📖 引用格式

若本项目对您的研究有所助益，请引用我们的论文：

@misc{zhao2026immersiongithubuniversescaling,
title={Immersion in the GitHub Universe: Scaling Coding Agents to Mastery},
author={Jiale Zhao and Guoxin Chen and Fanzhe Meng and Minghao Li and Jie Chen and Hui Xu and Yongshuai Sun and Xin Zhao and Ruihua Song and Yuan Zhang and Peng Wang and Cheng Chen and Jirong Wen and Kai Jia},
year={2026},
eprint={2602.09892},
archivePrefix={arXiv},
primaryClass={cs.SE},
url={https://arxiv.org/abs/2602.09892},
}

## 📄 许可证

本项目采用CC BY 4.0许可证进行授权——详细信息请参阅[LICENSE](LICENSE)文件。