allenai/c4

C4 Dataset Summary A colossal, cleaned version of Common Crawl's web crawl corpus. Based on Common Crawl dataset: "https://commoncrawl.org". This is the processed version of Google's C4 dataset We prepared five variants of the data: en, en.noclean, en.noblocklist, realnewslike, and multilingual (mC4). For reference, these are the sizes of the variants: en: 305GB en.noclean: 2.3TB en.noblocklist: 380GB realnewslike: 15GB multilingual (mC4): 9.7TB (108 subsets, one… See the full description on the dataset page: https://huggingface.co/datasets/allenai/c4.