dataseek/ptbr-blogs

PT-BR Blogs (long-form, C4-derived)是一个巴西葡萄牙语长文本博客数据集，包含185,000篇博客文章（每篇至少5,000词），从AllenAI C4 PT-BR切片中提取，通过筛选Blogspot、WordPress、Medium等博客平台域名获得。数据经过NFKD标准化和URL去重处理，用于巴西葡萄牙语语言模型的预训练、风格多样性和长上下文训练。数据集包含文本、URL、域名、时间戳和字符数等字段，未进行个人身份信息（PII）清洗，遵循ODC-BY许可。PT-BR Blogs (long-form, C4-derived) is a Brazilian Portuguese long-form blog dataset comprising 185,000 blog articles, each with a minimum of 5,000 words. It is extracted from the AllenAI C4 PT-BR slice, and curated by filtering domain names from blog platforms including Blogspot, WordPress, Medium and other similar platforms. The dataset has undergone NFKD normalization and URL deduplication, and is intended for pre-training, style diversity training and long-context training of Brazilian Portuguese language models. It contains fields such as text, URL, domain name, timestamp and character count. No Personally Identifiable Information (PII) scrubbing has been conducted, and the dataset is licensed under ODC-BY.