Multi-Agent Games (MGs)

该数据集包含了一系列随机生成的多代理游戏，旨在对ALMANAC算法进行评估，并与真实模型进行比较。数据集还包括了用于概率模型检测器PRISM的导出策略、游戏结构及规范说明。其规模根据状态空间的大小、代理数量以及规范数量的不同而有所变化，其任务是基准化多代理强化学习算法的性能。

This dataset contains a series of randomly generated multi-agent games, designed for evaluating the ALMANAC algorithm and making comparisons with real-world models. Furthermore, the dataset includes exported strategies, game structures and specification descriptions for the probabilistic model checker PRISM. The scale of the dataset varies according to the size of the state space, the number of agents and the number of specifications, and its primary task is to benchmark the performance of multi-agent reinforcement learning algorithms.