TW-Cooking

该数据集名为TW-Cooking，是一款关于收集各种烹饪食材并按照游戏内食谱准备餐点的游戏套装。它被用来验证EXPLORER方法的有效性。此外，这个数据集还是评估基于文本游戏中的神经符号代理的基准，涵盖了GATA数据集的前四个难度级别，任务聚焦于基于文本的游戏强化学习。This dataset, named TW-Cooking, is a game suite focused on collecting various cooking ingredients and preparing meals according to in-game recipes. It is utilized to validate the effectiveness of the EXPLORER method. Additionally, this dataset serves as a benchmark for evaluating neurosymbolic agents in text-based games, covering the first four difficulty levels of the GATA dataset, with tasks focused on reinforcement learning for text-based games.