$\\mathcal{D},_{\\textit{cnf},},$
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/AAIR-lab/auto-llm-assessment
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了400个不同的限制性从CNF公式，这些公式是通过改变一个比例参数生成的，旨在测试LLM模型在形式语法与自然语言之间翻译的有效性。该数据集是使用上下文无关文法即时生成的，非常适合用于评估LLM模型在形式翻译任务中的能力。规模达到了400个不同的CNF公式，任务则是实现形式语法（CNF）与自然语言之间的翻译。
This dataset includes 400 distinct restricted conjunctive normal form (CNF) formulas, which are generated by varying a scaling parameter. It is designed to evaluate the effectiveness of Large Language Models (LLMs) in translation between formal syntax and natural language. Generated on-the-fly using context-free grammars, this dataset is highly suitable for assessing LLMs' capabilities in formal translation tasks. The core task of this dataset is to perform translation between formal syntax (specifically CNF) and natural language.
提供机构：
AAIR lab
