Parkour Environment

该数据集专为测试Meta-ACL算法与深度强化学习学生们的可扩展性而设计，打造了一个全新的跑酷环境。该环境的特点是拥有一个被划分为400个平方单元的2D参数空间，用以模拟学生们不同的学习需求。在此规模上，我们设置了多个模拟环境，任务是为强化学习代理生成课程。

This dataset is specifically designed to evaluate the scalability of Meta-ACL algorithms and deep reinforcement learning learners. To support this evaluation, a novel parkour environment is constructed, which features a 2D parameter space partitioned into 400 square units to simulate the varied learning demands of the target learners. At this scale, multiple simulated environments are established, with the core task being the generation of curricula for reinforcement learning agents.