Skip to content
RAG Repo

CC-News is a dedicated news stream from the Common Crawl Foundation. Alongside its monthly web snapshots, Common Crawl continuously crawls news sites and packages the results as a separate dataset, so you get a steady flow of articles rather than a single point-in-time dump.

The data is published as WARC files (Web ARChive, the standard format that stores each crawled page's full HTTP response), the same format as the main Common Crawl archives. Because it is hosted on AWS as a public dataset, the cheapest way to work with it is from inside AWS, where there is no transfer cost. A ready-cleaned version is also mirrored on Hugging Face for quick loading.

For RAG, CC-News is a convenient source of news text for building current-affairs knowledge bases or fine-tuning models on journalistic writing. It has been used in several pretraining pipelines, including the corpus behind RoBERTa.

newsweb-crawlpretrainingenglishmultilingualcommon-crawl

Related sources