OpenSQZ/Classifiers-Data
收藏Hugging Face2025-11-23 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/OpenSQZ/Classifiers-Data
下载链接
链接失效反馈官方服务：
资源简介：
这是一个专门设计用于训练文本质量评估分类器的数据集，包含来自多个高质量语料库的注释数据，这些语料库涵盖了英语和中文的各种专业领域。数据集的总大小约为40B个token（经过采样），包含英语和中文两种语言，涉及通用文本、数学、编程、推理与问答等多个领域。注释维度包括数学智能和教育适宜性。数据集采用了基于LLM的自动化质量评估方法，并提供了详细的评分分布和质量标准。
This dataset is specifically designed for training text quality assessment classifiers, containing annotated data from multiple high-quality corpora covering both English and Chinese texts across various professional domains. It includes a total of ~40B tokens (after sampling), covers English and Chinese languages, and spans across General text, Mathematics, Programming, Reasoning & QA domains. The annotation dimensions include Mathematical intelligence and Educational suitability. The dataset employs an LLM-based automated quality assessment method and provides detailed score distribution and quality standards.
提供机构：
OpenSQZ
