---
license: apache-2.0
language:
- amh
- ara
- asm
- aym
- ben
- bul
- bzd
- cat
- cni
- deu
- ell
- eng
- ewe
- fra
- grn
- guj
- hau
- hch
- hin
- ibo
- ind
- jpn
- kan
- kin
- kor
- lin
- lug
- mal
- mar
- mya
- nah
- ori
- orm
- oto
- pan
- pat
- pol
- por
- quy
- ron
- rus
- shp
- sna
- sot
- spa
- swa
- tam
- tar
- tel
- tha
- tur
- twi
- urd
- vie
- wol
- xho
- yor
- zho
- zul

size_categories:
- n<1K

multilinguality:
- multilingual

pretty_name: globalnli

language_details: eng, amh, asm, aym, ara, ben, bul, bzd, cat, cni, deu, ell, ewe, fra, grn, guj, hau, hch, hin, ibo, ind, jpn, kan, kin, kor, lin, lug, mal, mar, mya, nah, ori, orm, oto, pan, pat, pol, por, quy, ron, rus, shp, sna, sot, spa, swa, tam, tar, tel, tha, tur, twi, urd, vie, wol, xho, yor, zho, zul

source_datasets:
- XNLI, AfriXNLI, IndicXNLI, AmericasNLI [30], XNLI-ca, myXNLI, IndoNLI, JNLI , InferBR, sick_pl, JamPatoisNLI, KLUE, RoNLI. 

tags:
- global_nli
- globalnli

task_categories:
- text-classification

task_ids:
- natural-language-inference

configs:
- config_name: amh
  data_files:
  - split: validation
    path: data/amh/dev.csv
  - split: test
    path: data/amh/test.csv
- config_name: ara
  data_files:
  - split: validation
    path: data/ara/dev.csv
  - split: test
    path: data/ara/test.csv
- config_name: asm
  data_files:
  - split: validation
    path: data/asm/dev.csv
  - split: test
    path: data/asm/test.csv
- config_name: aym
  data_files:
  - split: validation
    path: data/aym/dev.csv
  - split: test
    path: data/aym/test.csv
- config_name: ben
  data_files:
  - split: validation
    path: data/ben/dev.csv
  - split: test
    path: data/ben/test.csv
- config_name: bul
  data_files:
  - split: validation
    path: data/bul/dev.csv
  - split: test
    path: data/bul/test.csv
- config_name: bzd
  data_files:
  - split: validation
    path: data/bzd/dev.csv
  - split: test
    path: data/bzd/test.csv
- config_name: cat
  data_files:
  - split: validation
    path: data/cat/dev.csv
  - split: test
    path: data/cat/test.csv
- config_name: cni
  data_files:
  - split: validation
    path: data/cni/dev.csv
  - split: test
    path: data/cni/test.csv
- config_name: deu
  data_files:
  - split: validation
    path: data/deu/dev.csv
  - split: test
    path: data/deu/test.csv
- config_name: ell
  data_files:
  - split: validation
    path: data/ell/dev.csv
  - split: test
    path: data/ell/test.csv
- config_name: eng
  data_files:
  - split: validation
    path: data/eng/dev.csv
  - split: test
    path: data/eng/test.csv
- config_name: ewe
  data_files:
  - split: validation
    path: data/ewe/dev.csv
  - split: test
    path: data/ewe/test.csv
- config_name: fra
  data_files:
  - split: validation
    path: data/fra/dev.csv
  - split: test
    path: data/fra/test.csv
- config_name: grn
  data_files:
  - split: validation
    path: data/grn/dev.csv
  - split: test
    path: data/grn/test.csv
- config_name: guj
  data_files:
  - split: validation
    path: data/guj/dev.csv
  - split: test
    path: data/guj/test.csv
- config_name: hau
  data_files:
  - split: validation
    path: data/hau/dev.csv
  - split: test
    path: data/hau/test.csv
- config_name: hch
  data_files:
  - split: validation
    path: data/hch/dev.csv
  - split: test
    path: data/hch/test.csv
- config_name: hin
  data_files:
  - split: validation
    path: data/hin/dev.csv
  - split: test
    path: data/hin/test.csv
- config_name: ibo
  data_files:
  - split: validation
    path: data/ibo/dev.csv
  - split: test
    path: data/ibo/test.csv
- config_name: ind
  data_files:
  - split: validation
    path: data/ind/dev.csv
  - split: test
    path: data/ind/test.csv
- config_name: jpn
  data_files:
  - split: validation
    path: data/jpn/dev.csv
  - split: test
    path: data/jpn/test.csv
- config_name: kan
  data_files:
  - split: validation
    path: data/kan/dev.csv
  - split: test
    path: data/kan/test.csv
- config_name: kin
  data_files:
  - split: validation
    path: data/kin/dev.csv
  - split: test
    path: data/kin/test.csv
- config_name: kor
  data_files:
  - split: validation
    path: data/kor/dev.csv
  - split: test
    path: data/kor/test.csv
- config_name: lin
  data_files:
  - split: validation
    path: data/lin/dev.csv
  - split: test
    path: data/lin/test.csv
- config_name: lug
  data_files:
  - split: validation
    path: data/lug/dev.csv
  - split: test
    path: data/lug/test.csv
- config_name: mal
  data_files:
  - split: validation
    path: data/mal/dev.csv
  - split: test
    path: data/mal/test.csv
- config_name: mar
  data_files:
  - split: validation
    path: data/mar/dev.csv
  - split: test
    path: data/mar/test.csv
- config_name: mya
  data_files:
  - split: validation
    path: data/mya/dev.csv
  - split: test
    path: data/mya/test.csv
- config_name: nah
  data_files:
  - split: validation
    path: data/nah/dev.csv
  - split: test
    path: data/nah/test.csv
- config_name: ori
  data_files:
  - split: validation
    path: data/ori/dev.csv
  - split: test
    path: data/ori/test.csv
- config_name: orm
  data_files:
  - split: validation
    path: data/orm/dev.csv
  - split: test
    path: data/orm/test.csv
- config_name: oto
  data_files:
  - split: validation
    path: data/oto/dev.csv
  - split: test
    path: data/oto/test.csv
- config_name: pan
  data_files:
  - split: validation
    path: data/pan/dev.csv
  - split: test
    path: data/pan/test.csv
- config_name: pat
  data_files:
  - split: validation
    path: data/pat/dev.csv
  - split: test
    path: data/pat/test.csv
- config_name: pol
  data_files:
  - split: validation
    path: data/pol/dev.csv
  - split: test
    path: data/pol/test.csv
- config_name: por
  data_files:
  - split: validation
    path: data/por/dev.csv
  - split: test
    path: data/por/test.csv
- config_name: quy
  data_files:
  - split: validation
    path: data/quy/dev.csv
  - split: test
    path: data/quy/test.csv
- config_name: ron
  data_files:
  - split: validation
    path: data/ron/dev.csv
  - split: test
    path: data/ron/test.csv
- config_name: rus
  data_files:
  - split: validation
    path: data/rus/dev.csv
  - split: test
    path: data/rus/test.csv
- config_name: shp
  data_files:
  - split: validation
    path: data/shp/dev.csv
  - split: test
    path: data/shp/test.csv
- config_name: sna
  data_files:
  - split: validation
    path: data/sna/dev.csv
  - split: test
    path: data/sna/test.csv
- config_name: sot
  data_files:
  - split: validation
    path: data/sot/dev.csv
  - split: test
    path: data/sot/test.csv
- config_name: spa
  data_files:
  - split: validation
    path: data/spa/dev.csv
  - split: test
    path: data/spa/test.csv
- config_name: swa
  data_files:
  - split: validation
    path: data/swa/dev.csv
  - split: test
    path: data/swa/test.csv
- config_name: tam
  data_files:
  - split: validation
    path: data/tam/dev.csv
  - split: test
    path: data/tam/test.csv
- config_name: tar
  data_files:
  - split: validation
    path: data/tar/dev.csv
  - split: test
    path: data/tar/test.csv
- config_name: tel
  data_files:
  - split: validation
    path: data/tel/dev.csv
  - split: test
    path: data/tel/test.csv
- config_name: tha
  data_files:
  - split: validation
    path: data/tha/dev.csv
  - split: test
    path: data/tha/test.csv
- config_name: tur
  data_files:
  - split: validation
    path: data/tur/dev.csv
  - split: test
    path: data/tur/test.csv
- config_name: twi
  data_files:
  - split: validation
    path: data/twi/dev.csv
  - split: test
    path: data/twi/test.csv
- config_name: urd
  data_files:
  - split: validation
    path: data/urd/dev.csv
  - split: test
    path: data/urd/test.csv
- config_name: vie
  data_files:
  - split: validation
    path: data/vie/dev.csv
  - split: test
    path: data/vie/test.csv
- config_name: wol
  data_files:
  - split: validation
    path: data/wol/dev.csv
  - split: test
    path: data/wol/test.csv
- config_name: xho
  data_files:
  - split: validation
    path: data/xho/dev.csv
  - split: test
    path: data/xho/test.csv
- config_name: yor
  data_files:
  - split: validation
    path: data/yor/dev.csv
  - split: test
    path: data/yor/test.csv
- config_name: zho
  data_files:
  - split: validation
    path: data/zho/dev.csv
  - split: test
    path: data/zho/test.csv
- config_name: zul
  data_files:
  - split: validation
    path: data/zul/dev.csv
  - split: test
    path: data/zul/test.csv

---

# Dataset Card for global_nli

## Table of Contents
- [Table of Contents](#table-of-contents)
- [Dataset Description](#dataset-description)
  - [Dataset Summary](#dataset-summary)
  - [Languages](#languages)
- [Dataset Structure](#dataset-structure)
  - [Data Instances](#data-instances)
  - [Data Fields](#data-fields)
  - [Data Splits](#data-splits)
 
## Dataset Description

- **Point of Contact:** david.adelani@mila.quebec

### Dataset Summary

Global NLI is a new text-based benchmark based on the aggregation of existing NLI datasets that are publicly available.

### Languages

There are 59 languages available :

{
    'amh': 'Amharic',
    'ara': 'Arabic',
    'asm': 'Assamese',
    'aym': 'Aymara',
    'ben': 'Bengali',
    'bul': 'Bulgarian',
    'bzd': 'Bribri',
    'cat': 'Catalan',
    'cni': 'Asháninka',
    'deu': 'German',
    'ell': 'Greek',
    'eng': 'English',
    'ewe': 'Ewe',
    'fra': 'French',
    'grn': 'Guarani',
    'guj': 'Gujarati',
    'hau': 'Hausa',
    'hch': 'Wixarika',
    'hin': 'Hindi',
    'ibo': 'Igbo',
    'ind': 'Indonesian',
    'jpn': 'Japanese',
    'kan': 'Kannada',
    'kin': 'Kinyarwanda',
    'kor': 'Korean',
    'lin': 'Lingala',
    'lug': 'Luganda',
    'mal': 'Malayalam',
    'mar': 'Marathi',
    'mya': 'Burmese',
    'nah': 'Nahuatl',
    'ori': 'Odia (Oriya)',
    'orm': 'Oromo',
    'oto': 'Otomi',
    'pan': 'Punjabi',
    'pat': 'Jamaican Patois',
    'pol': 'Polish',
    'por': 'Portuguese',
    'quy': 'Quechua',
    'ron': 'Romanian',
    'rus': 'Russian',
    'shp': 'Shipibo-Conibo',
    'sna': 'chiShona',
    'sot': 'Sesotho',
    'spa': 'Spanish',
    'swa': 'Swahili',
    'tam': 'Tamil',
    'tar': 'Rarámuri',
    'tel': 'Telugu',
    'tha': 'Thai',
    'tur': 'Turkish',
    'twi': 'Twi',
    'urd': 'Urdu',
    'vie': 'Vietnamese',
    'wol': 'Wolof',
    'xho': 'isiXhosa',
    'yor': 'Yoruba',
    'zho': 'Chinese',
    'zul': 'isiZulu'
}

## Dataset Structure

### Data Instances

The examples look like this for English:

```
from datasets import load_dataset
data = load_dataset('McGill-NLP/GlobalNLI', 'eng') 
# Please, specify the language code
# A data point example is below:
{
'premise': 'The doors were locked when we went in.',
'hypothesis': 'All of the doors were open.',
'label': 0
}
```

### Data Fields

- `premise`: a multilingual string variable,
- `hypothesis`: a multilingual string variable,
- `label`: a classification label, with possible values including entailment (0), neutral (1), contradiction (2).

### Data Splits

All languages have two splits, `dev` and `test` set.

The splits have the following sizes :

| Language        | validation | test |
|-----------------|-----------:|-----:|
| English         |         30 |  600 |