SPoC

该数据集是一个用于微调二进制代码翻译任务模型的现实世界数据集，其中包含了来自编程竞赛平台的具有挑战性的程序。这些程序被编译成X86-64和ARM64二进制文件，并采用了从O0到O3不同级别的优化。该数据集的规模包括用于训练的3,294个程序和用于测试的364个程序，其任务专注于二进制代码翻译（Bct）。

This is a real-world dataset for fine-tuning models dedicated to binary code translation tasks. It includes challenging programs sourced from programming contest platforms, which are compiled into X86-64 and ARM64 binary files with optimization levels ranging from O0 to O3. The dataset comprises 3,294 training programs and 364 test programs, with its core task focused on binary code translation (Bct).