NetHack Dataset

该数据集是为了解决在NetHack环境中进行强化学习研究时，采用预先收集的数据集在离线强化学习中所面临的挑战而创建的大规模数据集。它包含了用于压缩游戏剧集的数据加载器，专为强化学习算法设计。尽管如此，实践者在实现和复现研究时可能会遇到一些困难。该数据集的规模达到了90GB，针对的任务是离线强化学习。

This large-scale dataset is developed to address the challenges inherent in offline reinforcement learning when using pre-collected datasets for reinforcement learning research conducted in the NetHack environment. It includes data loaders for compressed game episodes, which are specifically designed for reinforcement learning algorithms. Nonetheless, practitioners may encounter certain difficulties when implementing and reproducing their research. With a total size of 90 GB, this dataset is targeted at offline reinforcement learning tasks.