wiki40b

  • Description:

Clean-up text for 40+ Wikipedia languages editions of pages correspond to entities. The datasets have train/dev/test splits per language. The dataset is cleaned up by page filtering to remove disambiguation pages, redirect pages, deleted pages, and non-entity pages. Each example contains the wikidata id of the entity, and the full Wikipedia article after page processing that removes non-content sections and structured objects. The language models trained on this corpus - including 41 monolingual models, and 2 multilingual models - can be found at https://tfhub.dev/google/collections/wiki40b-lm/1

FeaturesDict({
    'text': Text(shape=(), dtype=string),
    'version_id': Text(shape=(), dtype=string),
    'wikidata_id': Text(shape=(), dtype=string),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
text Text string
version_id Text string
wikidata_id Text string
@inproceedings{49029,
title = {Wiki-40B: Multilingual Language Model Dataset},
author = {Mandy Guo and Zihang Dai and Denny Vrandecic and Rami Al-Rfou},
year = {2020},
booktitle   = {LREC 2020}
}

wiki40b/en (default config)

  • Config description: Wiki40B dataset for en.

  • Dataset size: 9.91 GiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 162,274
'train' 2,926,536
'validation' 163,597

wiki40b/ar

  • Config description: Wiki40B dataset for ar.

  • Dataset size: 833.20 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 12,271
'train' 220,885
'validation' 12,198

wiki40b/zh-cn

  • Config description: Wiki40B dataset for zh-cn.

  • Dataset size: 985.53 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 30,355
'train' 549,672
'validation' 30,299

wiki40b/zh-tw

  • Config description: Wiki40B dataset for zh-tw.

  • Dataset size: 986.45 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 30,670
'train' 552,031
'validation' 30,739

wiki40b/nl

  • Config description: Wiki40B dataset for nl.

  • Dataset size: 961.82 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 24,776
'train' 447,555
'validation' 25,201

wiki40b/fr

  • Config description: Wiki40B dataset for fr.

  • Dataset size: 3.37 GiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 68,004
'train' 1,227,206
'validation' 68,655

wiki40b/de

  • Config description: Wiki40B dataset for de.

  • Dataset size: 4.78 GiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 86,594
'train' 1,554,910
'validation' 86,068

wiki40b/it

  • Config description: Wiki40B dataset for it.

  • Dataset size: 2.00 GiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 40,443
'train' 732,609
'validation' 40,684

wiki40b/ja

  • Config description: Wiki40B dataset for ja.

  • Dataset size: 2.19 GiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 41,268
'train' 745,392
'validation' 41,576

wiki40b/ko

  • Config description: Wiki40B dataset for ko.

  • Dataset size: 453.98 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 10,802
'train' 194,977
'validation' 10,805

wiki40b/pl

  • Config description: Wiki40B dataset for pl.

  • Dataset size: 1.03 GiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 27,987
'train' 505,191
'validation' 28,310

wiki40b/pt

  • Config description: Wiki40B dataset for pt.

  • Dataset size: 1.08 GiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 22,693
'train' 406,507
'validation' 22,301

wiki40b/ru

  • Config description: Wiki40B dataset for ru.

  • Dataset size: 4.13 GiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 51,885
'train' 926,037
'validation' 51,287

wiki40b/es

  • Config description: Wiki40B dataset for es.

  • Dataset size: 2.70 GiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 48,764
'train' 872,541
'validation' 48,592

wiki40b/th

  • Config description: Wiki40B dataset for th.

  • Dataset size: 326.29 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 3,114
'train' 56,798
'validation' 3,093

wiki40b/tr

  • Config description: Wiki40B dataset for tr.

  • Dataset size: 308.87 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 7,890
'train' 142,576
'validation' 7,845

wiki40b/bg

  • Config description: Wiki40B dataset for bg.

  • Dataset size: 433.20 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 7,289
'train' 130,670
'validation' 7,259

wiki40b/ca

  • Config description: Wiki40B dataset for ca.

  • Dataset size: 753.00 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 15,568
'train' 277,313
'validation' 15,362

wiki40b/cs

  • Config description: Wiki40B dataset for cs.

  • Dataset size: 631.84 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 12,984
'train' 235,971
'validation' 13,096

wiki40b/da

  • Config description: Wiki40B dataset for da.

  • Dataset size: 240.51 MiB

  • Auto-cached (documentation): Yes (test, validation), Only when shuffle_files=False (train)

  • Splits:

Split Examples
'test' 6,219
'train' 109,486
'validation' 6,173

wiki40b/el

  • Config description: Wiki40B dataset for el.

  • Dataset size: 524.77 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 5,261
'train' 93,596
'validation' 5,130

wiki40b/et

  • Config description: Wiki40B dataset for et.

  • Dataset size: 184.07 MiB

  • Auto-cached (documentation): Yes (test, validation), Only when shuffle_files=False (train)

  • Splits:

Split Examples
'test' 6,205
'train' 114,464
'validation' 6,351

wiki40b/fa

  • Config description: Wiki40B dataset for fa.

  • Dataset size: 482.55 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 11,262
'train' 203,145
'validation' 11,180

wiki40b/fi

  • Config description: Wiki40B dataset for fi.

  • Dataset size: 534.13 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 14,179
'train' 255,822
'validation' 13,962

wiki40b/he

  • Config description: Wiki40B dataset for he.

  • Dataset size: 869.51 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 9,344
'train' 165,359
'validation' 9,231

wiki40b/hi

  • Config description: Wiki40B dataset for hi.

  • Dataset size: 277.56 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 2,643
'train' 45,737
'validation' 2,596

wiki40b/hr

  • Config description: Wiki40B dataset for hr.

  • Dataset size: 235.58 MiB

  • Auto-cached (documentation): Yes (test, validation), Only when shuffle_files=False (train)

  • Splits:

Split Examples
'test' 5,724
'train' 103,857
'validation' 5,792

wiki40b/hu

  • Config description: Wiki40B dataset for hu.

  • Dataset size: 634.25 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 15,258
'train' 273,248
'validation' 15,208

wiki40b/id

  • Config description: Wiki40B dataset for id.

  • Dataset size: 334.06 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 8,598
'train' 156,255
'validation' 8,714

wiki40b/lt

  • Config description: Wiki40B dataset for lt.

  • Dataset size: 140.46 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'test' 4,683
'train' 84,854
'validation' 4,754

wiki40b/lv

  • Config description: Wiki40B dataset for lv.

  • Dataset size: 80.07 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'test' 1,932
'train' 33,064
'validation' 1,857

wiki40b/ms

  • Config description: Wiki40B dataset for ms.

  • Dataset size: 142.49 MiB

  • Auto-cached (documentation): Yes (test, validation), Only when shuffle_files=False (train)

  • Splits:

Split Examples
'test' 5,235
'train' 97,509
'validation' 5,357

wiki40b/no

  • Config description: Wiki40B dataset for no.

  • Dataset size: 382.03 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 10,588
'train' 190,588
'validation' 10,547

wiki40b/ro

  • Config description: Wiki40B dataset for ro.

  • Dataset size: 319.68 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 7,870
'train' 139,615
'validation' 7,624

wiki40b/sk

  • Config description: Wiki40B dataset for sk.

  • Dataset size: 170.20 MiB

  • Auto-cached (documentation): Yes (test, validation), Only when shuffle_files=False (train)

  • Splits:

Split Examples
'test' 5,741
'train' 103,095
'validation' 5,604

wiki40b/sl

  • Config description: Wiki40B dataset for sl.

  • Dataset size: 157.38 MiB

  • Auto-cached (documentation): Yes (test, validation), Only when shuffle_files=False (train)

  • Splits:

Split Examples
'test' 3,341
'train' 60,927
'validation' 3,287

wiki40b/sr

  • Config description: Wiki40B dataset for sr.

  • Dataset size: 582.20 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 17,997
'train' 327,313
'validation' 18,100

wiki40b/sv

  • Config description: Wiki40B dataset for sv.

  • Dataset size: 613.62 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 22,291
'train' 400,742
'validation' 22,263

wiki40b/tl

  • Config description: Wiki40B dataset for tl.

  • Dataset size: 29.04 MiB

  • Auto-cached (documentation): Yes

  • Splits:

Split Examples
'test' 1,446
'train' 25,940
'validation' 1,472

wiki40b/uk

  • Config description: Wiki40B dataset for uk.

  • Dataset size: 1.67 GiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 26,581
'train' 477,618
'validation' 26,324

wiki40b/vi

  • Config description: Wiki40B dataset for vi.

  • Dataset size: 497.70 MiB

  • Auto-cached (documentation): No

  • Splits:

Split Examples
'test' 7,942
'train' 146,255
'validation' 8,195