Taxi1500 is a text classification dataset for evaluating the cross-lingual generalization ability of multilingual pre-trained language models. It introduces a sentence classification task with 6 topics and covers 1502 typologically diverse languages spanning 112 language families.
Dataloader name:
taxi1500_v2/taxi1500_v2.py
DataCatalogue: http://seacrowd.github.io/seacrowd-catalogue/card.html?taxi1500_v2