Competitive Multi-Agent Environments

该数据集包含了几项竞争性任务，在这些任务中，智能体在一个具有模拟物理效果的3D世界中相互竞争，通过自我对弈学习复杂的行为。数据集还提供了环境的代码和学习到的策略参数，这为复现研究结果和进一步的实验提供了可能。该数据集的规模涉及多个智能体在各种配置下的互动，其任务是进行多智能体强化学习。

This dataset encompasses several competitive tasks, wherein AI Agents compete against each other in a 3D world with simulated physics and learn complex behaviors through self-play. The dataset also provides the environment code and trained policy parameters, enabling the reproduction of research results and facilitating further experimental studies. The scale of this dataset involves interactions among multiple AI Agents under various configurations, and its core task focuses on multi-agent reinforcement learning.