Datasets

We've released hundreds of datasets in more than 75 different languages, both monolingual datasets and parallel ones. Processed and cleaned with own-developed or coauthored tools, ParaCrawl, Europat, MaCoCu or HPLT corpora series are part of our contributed dataset catalogue.

HPLT: High Performance Language Technologies

prompsit logo

+90 datasets

75 languages

Monolingual and bilingual

MaCoCu: Massive collection and curation of monolingual and bilingual data: focus on under-resourced languages

prompsit logo

+30 datasets

17 languages

Monolingual and bilingual

The Europat corpus: A parallel corpus of European patent data

prompsit logo

6 datasets

7 languages

Bilingual

ParaCrawl: Web-scale parallel corpora for the languages of the EU

prompsit logo

+40 datasets

56 languages

Bilingual + Monolingual (English)