Measurement System Datasets
收藏资源简介:
本数据集由德国维尔茨堡大学人工智能与数据科学中心的研究团队创建,旨在评估大型语言模型在不同测量系统中的泛化能力。数据集包含来自不同国家和地区、反映多元文化背景的财政数据、食品价格和城市距离信息。研究通过该数据集探讨了LLMs在默认测量系统选择、跨系统准确性和推理能力方面的表现,揭示了模型在不同文化背景下提供准确信息的能力,以及可能存在的偏见和局限性。
This dataset was developed by a research team from the Center for Artificial Intelligence and Data Science at the University of Würzburg, Germany, with the goal of evaluating the generalization capabilities of large language models (LLMs) across different measurement systems. It includes financial data, food prices, and urban distance information from various countries and regions, reflecting diverse cultural backgrounds. Using this dataset, the study explores the performance of LLMs in terms of default measurement system selection, cross-system accuracy, and reasoning abilities, revealing the models' capability to provide accurate information across different cultural contexts, as well as potential biases and limitations.

- 1On Generalization across Measurement Systems: LLMs Entail More Test-Time Compute for Underrepresented CulturesCenter For Artificial Intelligence and Data Science, University of Würzburg, Germany · 2025年



