Minesweeper Game Episodes

该数据集包含了专为训练使用多臂老虎机算法的智能体而设计的扫雷游戏各个回合的数据。具体来说，这些数据是从标准初学者级别的扫雷游戏中生成的，游戏面板大小为8x8，并设有10个地雷。该数据集的规模包括用于训练的百万回合以及用于测试的万次回合。任务旨在进行扫雷游戏的强化学习研究。

This dataset comprises data from individual gameplay rounds of the Minesweeper game, specifically created for training AI agents that utilize multi-armed bandit algorithms. These data are generated from standard beginner-level Minesweeper games, each featuring an 8×8 game board with 10 mines. The dataset includes one million rounds for training and 10,000 rounds for testing. It is designed to facilitate reinforcement learning research focused on the Minesweeper game.