pengyunie/codesearchnet-codegen

这是CodeSearchNet数据集的一个处理版本。具体来说，将`doc`（文档/文档字符串）、`sign`（函数签名）和`output`（函数体）分离为单独的字段，并将`doc`和`sign`按照编程语言的正确顺序连接成`problem`字段，使其适用于代码生成任务。在数据处理过程中，有731条记录由于各种原因被忽略。

This is a processed version of the CodeSearchNet dataset, specifically tailored for code generation tasks. The dataset separates the `doc` (documentation/docstring), `sign` (function signature), and `output` (function body) into distinct fields; these fields are then concatenated into a `problem` field, which is suitable for code generation tasks. During processing of the original dataset, 731 records were ignored due to various issues such as missing docstrings, parsing errors, and code cutoff problems across different programming languages. The dataset is multilingual in nature, focusing on code as the primary language, and is licensed under the Apache 2.0 license.