人工标注的大型语言模型生成的文本数据集
收藏资源简介:
该数据集是由上海交通大学网络安全科学与工程学院的研究人员创建的,包含由大型语言模型生成的文本和人类生成的文本。数据集大小为200条文本,旨在用于评估和改进大型语言模型检测和解释其生成文本的能力。数据集的创建过程涉及选择现有的人类生成的文本,设计相应的提示语以生成大型语言模型文本,然后由人工标注者对文本来源进行判断并提供解释。该数据集的应用领域是自然语言处理,特别是用于解决区分大型语言模型生成的文本和人类生成的文本这一难题。
This dataset was developed by researchers from the School of Cyberspace Science and Engineering, Shanghai Jiao Tong University. It contains two types of texts: those generated by large language models (LLMs) and human-written texts, with a total of 200 text samples. This dataset is designed to evaluate and enhance the capability of large language models to detect their generated texts and provide corresponding explanatory justifications. The dataset construction process involves three key steps: first, selecting existing human-generated texts; second, crafting corresponding prompts to generate LLM-based texts; third, having human annotators judge the source of each text and provide their reasoning for the judgment. The application domain of this dataset is natural language processing (NLP), specifically targeting the challenging task of distinguishing between texts generated by large language models and those created by human authors.




