TabMWP-TeLL
收藏资源简介:
TabMWP-TeLL数据集是由西安交通利物浦大学和利物浦大学的研究团队基于TabMWP数据集扩展创建的高质量表格数学应用题数据集。该数据集包含38,431条问题,涵盖多种表格背景和数学逻辑,旨在通过模板驱动和LLM重述的方法生成多样化和正确性高的数学应用题。数据集的创建过程包括从现有样本中提取模板、使用LLM扩展和重述问题,并添加详细的推理步骤。该数据集主要应用于评估和提升大语言模型在数学推理任务中的表现,特别是在解决复杂表格数学应用题方面。
The TabMWP-TeLL dataset is a high-quality tabular math word problem dataset extended from the original TabMWP dataset, developed by a research team from Xi'an Jiaotong-Liverpool University and the University of Liverpool. This dataset contains 38,431 problem entries, covering diverse tabular backgrounds and mathematical logics, and is designed to generate diverse and highly accurate math word problems through template-driven and LLM-based paraphrasing approaches. The dataset creation workflow includes extracting templates from existing samples, expanding and paraphrasing problems using LLMs, and adding detailed reasoning steps. This dataset is primarily used to evaluate and enhance the performance of large language models (LLMs) in mathematical reasoning tasks, particularly in solving complex tabular math word problems.

- 1Template-Driven LLM-Paraphrased Framework for Tabular Math Word Problem Generation西安交通利物浦大学, 利物浦大学, 杜克昆山大学 · 2024年



