D4RL Offline RL Datasets

该数据集是由D4RL收集的离线日志数据集，旨在评估离线强化学习算法。该数据集涵盖了从HalfCheetah、Hopper和Walker2d环境中不同策略收集的四个离线日志数据集：随机、中等、中等回放和中等专家。这些数据集具有不同策略下的多尺度性能水平。其任务是针对离线强化学习。

This dataset is an offline reinforcement learning log dataset collected by D4RL, designed to evaluate offline reinforcement learning algorithms. It includes four offline log datasets gathered via different policies across the HalfCheetah, Hopper, and Walker2d environments: random, medium, medium replay, and medium expert. These datasets exhibit multi-scale performance levels corresponding to distinct policies. This dataset is tailored for offline reinforcement learning tasks.