Fùxì
收藏资源简介:
Fùxì是一个全面评估语言模型在古代中文文本理解和生成方面的基准,由华东师范大学计算机科学与技术学院创建。该数据集包含21种不同类型的任务,涵盖了理解和生成两个方面,特别设计的任务类型能评估模型在古代中文方面的独特能力,如诗歌创作、对子完成和文体转换等。数据集内容来源于文学、哲学、历史等多个领域,经过精心筛选和整理,旨在促进古代中文文本处理的研究和发展。
Fùxì is a benchmark for comprehensively evaluating language models' understanding and generation capabilities of ancient Chinese texts, developed by the School of Computer Science and Technology, East China Normal University. This dataset encompasses 21 distinct task types covering both understanding and generation aspects. The specially designed task categories are intended to evaluate the unique capabilities of models in ancient Chinese processing, such as poetry composition, couplet completion, and style transfer. The dataset's content is sourced from multiple fields including literature, philosophy, history and other disciplines, and has undergone rigorous screening and curation, with the aim of promoting research and development in ancient Chinese text processing.

- 1Fùxì: A Benchmark for Evaluating Language Models on Ancient Chinese Text Understanding and Generation华东师范大学计算机科学与技术学院 · 2025年



