AIDER

该数据集名为Aider，包含133个Python编程练习题，每道题目都附带自然语言描述和一系列单元测试。该数据集旨在评估参与者将自然语言任务描述翻译成可执行代码的能力，这些代码必须通过一系列预先定义的单元测试。规模上，数据集共有133个练习题，任务内容涉及简单的函数级别代码生成。This dataset, named Aider, contains 133 Python programming exercises. Each exercise comes with a natural language description and a set of unit tests. The dataset aims to evaluate participants' ability to translate natural language task descriptions into executable code that must pass a series of predefined unit tests. In terms of scale, the dataset has 133 exercises, and the task content covers simple function-level code generation.