mlfoundations-dev/evalchemy_swebench_oracle

该数据集包含了代码相关的信息和元数据，如实例ID、文本内容、仓库信息、基础提交信息、问题陈述、提示文本、创建时间、补丁信息、测试补丁、版本信息以及两种转换状态（失败转成功和成功转成功）。此外，还包含了环境设置提交信息、上下文信息（内容与角色）以及生成参数（是否抽样、最大新令牌数、种子和温度）。数据集被划分为测试集，包含了2294个示例，总大小为453,618,339字节。

The dataset includes code-related information and metadata such as instance ID, text content, repository information, base commit, problem statement, hint text, creation time, patch information, test patch, version information, and two transition states (failure to pass and pass to pass). It also includes environment setup commit information, context information (content and role), and generation parameters (whether to sample, maximum number of new tokens, seed, and temperature). The dataset is split into a test set, containing 2294 examples with a total size of 453,618,339 bytes.