hkust-nlp/CodeIO-PyEdu-Reasoning-Raw

CodeI/O数据集是一个用于通过代码输入输出预测来压缩推理模式的数据集。它包含了处理过的PythonEdu-Reasoning数据集的原始数据，每条数据包括问题描述、输入输出要求与约束、参考代码、函数名称、输入输出对、代码来源、推理类型和元信息等。该数据集的部分输入输出对可能为空，因为它们在执行时超出了大小限制。

The CodeI/O dataset is designed for condensing reasoning patterns through code input-output prediction. It contains the raw data of the processed PythonEdu-Reasoning dataset, with each entry including a problem description, input/output requirements and constraints, reference code, function name, input/output pairs, source of the raw code files, reasoning type, and metadata. Some of the input/output pairs may be empty due to exceeding the size constraints during execution.