Vidhi03/HC3
收藏Hugging Face2026-04-27 更新2026-05-03 收录
下载链接：
https://hf-mirror.com/datasets/Vidhi03/HC3
下载链接
链接失效反馈官方服务：
资源简介：
HC3数据集是第一个人类与ChatGPT生成的文本比较语料库，旨在评估ChatGPT与人类专家的接近程度。该数据集支持多种自然语言处理任务，包括文本分类、问答、句子相似性和零样本分类，涵盖英文和中文两种语言。数据集标签涉及ChatGPT、SimpleAI、检测和OOD等，规模在10K到100K之间。相关论文和代码可在提供的GitHub链接中找到。
The HC3 dataset is the first human-ChatGPT comparison corpus, designed to evaluate how close ChatGPT is to human experts. It supports various NLP tasks including text-classification, question-answering, sentence-similarity, and zero-shot-classification, and covers both English and Chinese languages. The dataset is tagged with ChatGPT, SimpleAI, Detection, and OOD, and its size ranges between 10K and 100K. The related paper and code are available via the provided GitHub link.
提供机构：
Vidhi03
