mlfoundations-dev/a1_code_kodcode

这是一个包含代码生成任务的数据集，其中包括问题的版本、风格、子集等信息。数据集的每个示例都包含了问题ID、解决方案、测试代码及其结果等信息。此外，还包含了GPT模型生成代码的尝试次数、难度、通过率等信息。每个尝试都详细记录了代码的来源、测试覆盖率等。数据集还包含了原始指令、提示ID等元数据信息。整个数据集以训练集的形式提供，包含了约3.99万个示例。

This is a dataset containing code generation tasks, including information such as version, style, subset of the problem. Each example in the dataset contains the problem ID, solution, test code and its results, etc. In addition, it includes information such as the number of attempts, difficulty, and pass rate of the GPT model in generating code. Each attempt is detailed with the source of the code, test coverage, etc. The dataset also contains metadata information such as the original instruction, prompt ID, etc. The entire dataset is provided in the form of a training set, containing about 39,983 examples.