SolarisCipher/hk_content_corpus
收藏Hugging Face2025-08-16 更新2025-10-25 收录
下载链接：
https://hf-mirror.com/datasets/SolarisCipher/hk_content_corpus
下载链接
链接失效反馈官方服务：
资源简介：
HK内容语料库（粤语和繁体中文）包含来自公开网络来源的八个特定源清洁文本语料库，包括新闻网站、在线论坛、百科全书和餐厅评论。该数据集最初用于研究双言制对香港语言建模影响的研究。每个文件存储纯UTF-8文本，每条记录占一行，空行作为分隔符。数据集可用于语言模型预训练和微调，粤语/繁体中文语言建模，以及下游任务如分类或生成。
The HK Content Corpus (Cantonese & Traditional Chinese) contains eight source-specific cleaned text corpora from public web sources, including news sites, online forums, encyclopedia, and restaurant reviews. Initially created for research on the effect of diglossia on Hong Kong language modeling, each file stores plain UTF-8 text with one record per line and blank lines as separators. The dataset is suitable for language model pretraining and finetuning, Cantonese/Traditional Chinese linguistic modeling, and downstream tasks such as classification or generation.
提供机构：
SolarisCipher
