Skip to content

Dataset · facebook

XNLI: Textual entailment in 14 languages

facebook/xnli

A translated portion of MNLI for predicting logical entailment, contradiction, or neutrality between two phrases.

What's inside

XNLI is a subset of several thousand MNLI examples translated into 14 languages, including lower-resource ones. The task is to predict whether the second sentence follows from the first, contradicts it, or is neutral. This is a three-label sentence pair classification task.

What you can do with it

  • Training a textual entailment model in multiple languages
  • Testing cross-lingual transfer
  • Evaluating multilingual LLMs
Languages
Arabic, Bulgarian, German, Greek, English, Spanish, French, Hindi and 4 more
Size
1M–10M rows
Downloads this month
59.9K
Likes
76

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("facebook/xnli")