Atari 100k

该数据集是由Arcade学习环境（ALE）提供的框架，旨在开发和评估一系列Atari 2600游戏的强化学习算法。该数据集包含了多种游戏，用于测试在带有和不含参数噪声的情况下DQN的性能。这些游戏的复杂度各不相同，共计21款，构成了用于离散动作环境的强化学习基准测试任务。

This dataset is a framework provided by the Arcade Learning Environment (ALE), dedicated to developing and evaluating reinforcement learning algorithms for a series of Atari 2600 games. It includes various games to test the performance of Deep Q-Networks (DQN) both with and without parameter noise. Comprising 21 games with varying complexity levels, this dataset constitutes a benchmark test suite for reinforcement learning tasks in discrete action environments.