This dataset gathers 728,321 biographies from wikipedia. It aims at evaluating text generation algorithms. For each article, we provide the first paragraph and the infobox (both tokenized). - View it on GitHub