SWE-rebench nebius
Dataset Summary SWE-rebench is a large-scale dataset designed to support training and evaluation of LLM-based software engineering (SWE) agents, building upon and expanding our earlier release, SWE-bench-extra. It is constructed using a fully automated pipeline that continuously extracts real-world interactive SWE tasks from GitHub repositories at scale, as detailed in our paper SWE-rebench: An Automated Pipeline for Task Collection and Decontaminated Evaluation of Software… See the full description on the dataset page: https://huggingface.co/datasets/nebius/SWE-rebench.
- 类型
- dataset
- 许可
- cc-by-4.0
- 下载量
- 382,468
- 点赞
- 71
- 访问
- public
- 文件
- 0
标签
- 代码数据
- 评测基准
- Agent
- 代码生成
- 大语言模型
- 数据处理
- 深度学习
摘要
SWE-rebench是一个大规模软件工程（SWE）智能体训练与评测数据集，基于全自动流水线从GitHub持续提取真实交互式SWE任务，覆盖3400多个Python仓库、超过2.1万个issue-PR配对。它扩展了SWE-bench方法论，增加LLM驱动的环境安装指令提取和任务质量评估，并提供预构建Docker镜像以简化环境搭建。适用于LLM驱动的软件工程智能体（如SWE-agent）的训练、评估与排行榜构建。
README
--- license: cc-by-4.0 task_categories: - other library_name: datasets dataset_info: features: - name: instance_id dtype: string - name: base_commit dtype: string - name: created_at dtype: string - name: environment_setup_commit dtype: string - name: hints_text dtype: string - name: patch dtype: st…