Is this your dataset? Claim it with the Hugging Face account that owns it.

allenai/
c4

Name: c4
Creator: allenai
License: https://spdx.org/licenses/ODC-By-1.0.html

text-generation

fill-mask

language-modeling

masked-language-modeling

Views

No views yet

	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.