OSCAR is currently shuffled at line level and no metadata is provided. Thus it is mainly intended to be used in the training of unsupervised language models for NLP.
Data is distributed by language in both original and deduplicated form. There are currently 166 different languages available. If you use OSCAR please consider giving us some feedback using the contact form down below. Also consider citing our paper.
If you want to contribute to OSCAR, for example by tokenizing one of the corpora for a particular language, or by helping us translate our webpage, please open a pull request here.