TCELongBench
收藏arXiv2024-06-05 更新2024-08-06 收录
下载链接：
http://arxiv.org/abs/2406.02472v1
下载链接
链接失效反馈官方服务：
资源简介：
TCELongBench是由复旦大学计算机科学学院创建的大型数据集，专注于评估大型语言模型在处理时间复杂事件和长文本理解方面的能力。该数据集包含2,289个时间复杂事件（TCE），总计88,821个问答对，涉及阅读理解、时间序列理解和未来事件预测三个任务。数据集通过生成-验证范式确保高质量，旨在帮助模型更好地理解和预测复杂事件的发展。
TCELongBench is a large-scale dataset developed by the School of Computer Science, Fudan University, which focuses on evaluating the capabilities of large language models (LLMs) in handling temporally complex events and long-text understanding. This dataset includes 2,289 temporally complex event (TCE) instances, with a total of 88,821 question-answer pairs, covering three tasks: reading comprehension, time series understanding, and future event prediction. It ensures high data quality via a generation-verification paradigm, and aims to assist models in better understanding and predicting the evolution of complex events.
提供机构：
复旦大学计算机科学学院
创建时间：
2024-06-05
