遇见数据集

SWE-Gym/SWE-gym_Lite

收藏
Hugging Face2024-12-11 更新2024-12-14 收录
官方服务:

资源简介:

--- dataset_info: features: - name: instance_id dtype: string - name: hints_text dtype: string - name: patch dtype: string - name: test_patch dtype: string - name: created_at dtype: string - name: problem_statement dtype: string - name: repo dtype: string - name: base_commit dtype: string - name: version dtype: string - name: PASS_TO_PASS sequence: string - name: FAIL_TO_PASS sequence: string splits: - name: train num_bytes: 3389442 num_examples: 230 download_size: 931193 dataset_size: 3389442 configs: - config_name: default data_files: - split: train path: data/train-* --- SWE-Gym Lite contains 230 instances sourced from 11 Python repos, following SWE-Bench Lite data collection procedure. **Get started at project page [swe-gym.github.io](swe-gym.github.io)** ![image/png](https://cdn-uploads.huggingface.co/production/uploads/61568f37272f2d87a99ba884/HEq_gwToxKkOAaBrIpBBe.png) ![image/png](https://cdn-uploads.huggingface.co/production/uploads/61568f37272f2d87a99ba884/9GtLvWc4ntFoc--NQQg7x.png)

SWE-Gym Lite contains 230 instances sourced from 11 Python repos, following SWE-Bench Lite data collection procedure.

提供机构:
SWE-Gym
搜集汇总
数据集介绍
SWE-Gym/SWE-gym_Lite 数据集图片
背景与挑战
背景概述
SWE-Gym-Lite是一个轻量级软件工程任务数据集,包含230个来自11个Python仓库的实例,遵循SWE-Bench Lite的收集流程。每个实例提供问题描述、补丁、测试等信息,旨在评估和提升代码生成与修复能力。
以上内容由遇见数据集搜集并总结生成
二维码
社区交流群
二维码
科研交流群
商业服务