Atari 2600 game Montezuma’s Revenge

该数据集包含了用于训练强化学习代理在Atari 2600游戏《蒙特祖玛的复仇》中导航并达成特定子目标的演示轨迹。此外，该数据集还用于为学习代理识别抽象的子目标，使得代理能够在低层次和高层次上高效学习。该数据集规模为10条演示轨迹，相关任务是强化学习。

This dataset contains demonstration trajectories used for training reinforcement learning (RL) agents to navigate and complete specific sub-goals in the Atari 2600 game *Montezuma's Revenge*. Additionally, this dataset is also employed to assist learning agents in identifying abstract sub-goals, enabling efficient learning at both low-level and high-level levels. This dataset comprises 10 demonstration trajectories, and its associated task falls under the domain of reinforcement learning.