Decima

该数据集名为Decima，是一种基于最新图神经网络（GNN）技术的作业调度器，用于验证单步和多步策略证明性质。该数据集包含从Decima原生测试床中抽取的初始作业状态，由平均每个作业9.2个节点和8.5条边的稀疏图组成。具体规模方面，包含5个或10个作业的作业配置文件，其对应的激活中位数分别为5845和10997。该数据集的任务是验证基于GNN的作业调度器。

This dataset, named Decima, is a job scheduler leveraging state-of-the-art Graph Neural Networks (GNNs), developed to verify the provable properties of single-step and multi-step scheduling policies. It comprises initial job states sampled from the native Decima testbed, which are structured as sparse graphs with an average of 9.2 nodes and 8.5 edges per individual job. Regarding specific scales, the dataset includes job configuration profiles with 5 or 10 jobs, whose corresponding median activation counts are 5845 and 10997 respectively. The core task of this dataset is to verify GNN-based job schedulers.