Datasets

By now, CoBaLD suggests manually annotated datasets for English and Russian consisting of more than 180 000 and 400 000 tokens correspondingly, Hungarian and Serbian datasets about 7-8 thousand tokens each which were obtained through the cross-lingual transfer (CLT) technique and checked manually afterwards. Moreover, we have several small datasets for Spanish, German, French, Japanese, Romanian, and Swedish, which consist of manually checked sentences as well and work on creating larger datasets for the languages with the help of CLT.