TextGames

该数据集名为TextGames，旨在通过挑战性的基于文本的游戏来评估大型语言模型（LLM）的能力，这些游戏需要高级技能，如模式识别、空间感知、算术和逻辑推理。该数据集允许比较人类与LLM的能力，记录的指标包括解决率、尝试次数以及解决问题所需的时间。它包含了单轮次和多轮次推理任务，并已对多种LLM进行了测试，包括开源和专有模型。其任务是评估LLM在解决基于文本的谜题游戏中的表现。

This dataset, named TextGames, is designed to evaluate the capabilities of Large Language Models (LLMs) via challenging text-based games that require advanced skills including pattern recognition, spatial awareness, arithmetic, and logical reasoning. It enables the comparison of performance between humans and LLMs, with recorded metrics such as solve rate, number of attempts, and time required to solve the given problems. The dataset encompasses both single-turn and multi-turn reasoning tasks, and has been tested on a variety of LLMs including both open-source and proprietary models. The core task of this dataset is to assess the performance of LLMs when solving text-based puzzle games.