Easily convert common crawl to a dataset of caption and document. Image/text Audio/text Video/text, ... - View it on GitHub