metadata
license: apache-2.0
This dataset contains ~200K wiki articles, randomly sampled from the wiki dataset. The articles have been tokenized using AutoTokenizer.from_pretrained('bert-base-uncased'). The dataset was created for the research porpouses.