Test Set I
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/IpadLi/Grasper
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是从训练集中抽取的30个游戏样本，用于评估在分布内的表现。此外，该测试集还用于评估Grasper的零样本性能。规模为30个游戏，任务是对在分布游戏上Grasper的性能进行评估。
This dataset consists of 30 game samples extracted from the training set, designed to evaluate in-distribution performance. Furthermore, this test set is also employed to assess the zero-shot performance of Grasper. Containing 30 games in total, its core task is to evaluate Grasper's performance on in-distribution games.
提供机构：
Authors of the paper
