HiTZ/GuideX_pre-training_data
收藏Hugging Face2025-06-12 更新2025-07-05 收录
下载链接：
https://hf-mirror.com/datasets/HiTZ/GuideX_pre-training_data
下载链接
链接失效反馈官方服务：
资源简介：
GUIDEX数据集是一个大规模的合成生成资源，用于训练信息提取模型。它由来自FineWeb-edu数据集的大约10,000个文档组成，每个文档都使用LLama 3.1-70B-Instruct模型和GUIDEX管道自动注释。数据集具有高度多样性，包含28,677个独特的实体标签，涵盖医学、经济学、历史、音乐和教育等领域。它以与现代IE框架兼容的代码样式格式化。
The GUIDEX dataset is a large-scale, synthetically generated resource for training Information Extraction models. It consists of approximately 10,000 documents sampled from the FineWeb-edu dataset, each annotated automatically using the Llama 3.1-70B-Instruct model guided by the GUIDEX pipeline. The dataset is highly diverse, containing 28,677 unique entity labels across domains such as Medicine, Economics, History, Music, and Education. It is formatted in a code-style compatible with modern IE frameworks.
提供机构：
HiTZ
