eth-sri/SWT-bench_Lite_bm25_27k_zsb

SWT-bench Lite是一个SWT-bench数据集的子集，旨在测试系统自动重现GitHub问题的能力。该数据集包含了从11个流行的Python GitHub项目中收集的276个测试Issue-Pull Request对，用于评估模型在重现和验证真实世界中的错误修复方面的表现。SWT-bench Lite is a subset of the SWT-bench dataset, designed to test the ability of systems to automatically reproduce GitHub issues. The dataset consists of 276 test Issue-Pull Request pairs collected from 11 popular Python GitHub projects, used to evaluate the performance of models in reproducing and validating real-world bug fixes.