CIRCLE

该数据集是一组多任务控制任务，被广泛应用于元强化学习的经典基准测试中。该数据集包含了各种任务的专业轨迹，这些轨迹被用来评估不同元强化学习算法的表现。这些任务规模多样，是从任务分布中抽取的，旨在研究具有少样本泛化和任务适应能力的元强化学习。This dataset consists of a set of multi-task control tasks, which are widely utilized as a classic benchmark in meta-reinforcement learning research. It contains expert trajectories for diverse tasks, and these trajectories are used to evaluate the performance of various meta-reinforcement learning algorithms. These tasks, which are sampled from a task distribution and vary in scale, are designed to study meta-reinforcement learning approaches that possess few-shot generalization and task adaptation capabilities.