Climb Game

该数据集是对攀登游戏的时间扩展版本，旨在评估多种病理环境下多代理强化学习算法的性能。该数据集包含了1600万状态-动作对，用于评估包括宽容性和滞后Q学习等策略在内的多种多代理强化学习算法的表现。此外，该数据集的任务专注于多代理强化学习。

This dataset is a temporally extended version of the Climb Game, designed to evaluate the performance of multi-agent reinforcement learning algorithms across various pathological environments. It contains 16 million state-action pairs, which are used to assess the performance of a variety of multi-agent reinforcement learning algorithms including strategies such as Tolerant Q-learning and Lagged Q-learning. Furthermore, the tasks included in this dataset focus on multi-agent reinforcement learning.