parsaidp/SWE-bench_Verified_easy

该数据集包含194个测试样本，专注于代码补全或测试修复任务，涉及GitHub仓库的代码变更。特征包括仓库名称、实例ID、基础提交、代码补丁、测试补丁、问题陈述、提示文本、创建时间、版本、失败到通过和通过到通过的测试结果、环境设置提交和难度级别。数据集用于评估或训练模型在代码生成和测试修复方面的性能。This dataset comprises 194 test samples focused on code completion or test repair tasks related to code changes in GitHub repositories. Its features include repository name, instance ID, base commit, code patch, test patch, problem statement, prompt text, creation time, version, test results of transitioning from failure to pass and from pass to pass, environment setup commit, and difficulty level. This dataset is used to evaluate or train models' performance on code generation and test repair.