Arko007/ultimate-fake-news-dataset

终极假新闻数据集是一个大规模的英文二分类数据集，包含约925万个经过筛选和去重的样本，主要用于训练文本分类模型，识别长篇新闻文章和标题中的假新闻。数据集整合了多个公开的假新闻语料库、精选出版商内容和Kaggle数据集，支持领域自适应预训练和大规模微调。

The Ultimate Fake News Dataset is a large-scale English binary classification dataset containing approximately 9.25 million curated and deduplicated samples, primarily intended for training text-classification models to detect fake news in long-form news articles and headlines. The dataset aggregates multiple public fake-news corpora, selected publisher content, and Kaggle datasets to support domain-adaptive pretraining and large-scale fine-tuning.