AgarCL

该数据集名为AgarCL，是一个基于Agar.io游戏的持续强化学习研究平台，具有随机动力学、连续动作和部分可观测性等特点。它包含了一系列迷你游戏，这些游戏旨在隔离特定的环境挑战，并评估不同的强化学习算法。该数据集规模庞大，拥有多个迷你游戏环境，其任务是评估代理在持续强化学习中的学习能力。

This dataset, named AgarCL, is a continuous reinforcement learning research platform based on the Agar.io game, featuring stochastic dynamics, continuous action spaces, and partial observability. It encompasses a suite of mini-games developed to isolate specific environmental challenges and evaluate various reinforcement learning algorithms. Featuring a substantial scale with multiple mini-game environments, this dataset is intended to assess the learning capabilities of agents in continuous reinforcement learning.