E1 Programs

该数据集包含了105036个训练程序，每个程序的长度不超过4，且每个程序包含五个输入输出示例对。在数据生成过程中，确保了训练程序与测试程序之间没有重叠，且与较短的或等长度的程序在功能上不等效。该数据集的规模为105036个训练程序，任务是对示例进行程序合成。

This dataset contains 105,036 training programs, each with a maximum length of 4 and including five input-output example pairs. During the dataset generation process, it is ensured that there is no overlap between training programs and test programs, and that each training program is not functionally equivalent to any program with a shorter length or the same length. The dataset comprises 105,036 training programs, and its core task is program synthesis from input-output examples.