Natural Language Inference (NLI) datasets in Czech

The goal of the Natural Language Inference (NLI) task is to determine whether a "hypothesis" is true (entailment), false (contradiction), or undetermined (neutral) given a "premise". The repository contains three NLI datasets, namely snli (https://huggingface.co/datasets/snli), multi_nli (https://huggingface.co/datasets/multi_nli) and qnli (https://huggingface.co/datasets/glue/viewer/qnli/train). These datasets are in two versions, the original English version and our-added Czech translation using CUBBITT, the Charles University Block-Backtranslation-Improved Transformer Translation model (https://lindat.mff.cuni.cz/services/translation/). The record includes target labels for Czech datasets as well, however, note that they could no longer be correct for the Czech translation (because of errors made by the translation model). The licence of this record (CC BY-SA) holds for the translated part of the dataset. For the original English datasets, follow their respective licence descriptions.