Grid Soccer

该数据集包含了在网格足球环境中训练和评估多代理协议的内容。在这个环境中，代理们学习如何相互合作与协调，以对抗一个手工编码的策略，目的是进球。此外，有关参数共享团队与协调学习团队对战的视频可以在以下链接中找到：https://drive.google.com/open?id=1jntEENSbnW_oyoFZJuCobBHybO_jdpUy。该数据集的规模涉及数十万次的训练迭代，其任务是在网格足球模拟中实现多代理协作与强化学习。

This dataset is designed for training and evaluating multi-agent protocols within a grid-based football simulation environment. In this setup, agents learn to cooperate and coordinate with one another to compete against a hand-coded opposing strategy with the core objective of scoring goals. Additionally, videos showcasing matches between parameter-sharing teams and coordinated learning teams are accessible via the following link: https://drive.google.com/open?id=1jntEENSbnW_oyoFZJuCobBHybO_jdpUy. Spanning hundreds of thousands of training iterations, this dataset targets multi-agent collaboration and reinforcement learning tasks in grid football simulations.