MAgent

该数据集是由MAgent框架提供的，它为强化学习封装了各种大规模的多智能体环境。在这些环境中，智能体能够在复杂的场景中进行互动和学习。MAgent不仅提供了多样化的设置和场景以评估多智能体强化学习算法，还包括了合作与竞争的动态机制。其规模之大，在例如Gather环境中，可以同时存在数百个智能体和数千个食物碎片。该数据集的任务主要集中在多智能体强化学习领域。

This dataset is provided by the MAgent framework, which encapsulates various large-scale multi-agent environments for reinforcement learning. In these environments, agents can interact and learn within complex scenarios. MAgent not only offers diverse configurations and scenarios for evaluating multi-agent reinforcement learning algorithms, but also incorporates cooperative and competitive dynamic mechanisms. Notably, the scale of these environments is considerable: for example, in the Gather environment, hundreds of agents and thousands of food fragments can coexist simultaneously. The tasks of this dataset are primarily focused on the field of multi-agent reinforcement learning.