lhoestq HF staff commited on
Commit
2c78958
1 Parent(s): 607bd4c

🤗 Datasets support

Browse files

Finally took the time to configure the dataset to be loaded in 🤗 Datasets.
This will also add a dropdowns to the Dataset Viewer with "en", "en.noclean", "en.noblocklist", "realsnewslike", "multilingual", as well as each mC4 language "af", "ar", "ar", etc.

Bonus: I documented how to load it not only with 🤗 Datasets but also with Dask.

Ultimately I think we can redirect the [c4](https://huggingface.co/datasets/c4) and [mc4](https://huggingface.co/datasets/mc4) pages to this one

Files changed (1) hide show
  1. README.md +1355 -8
README.md CHANGED
@@ -1,6 +1,1121 @@
1
- This is the processed version of [Google's C4 dataset](https://www.tensorflow.org/datasets/catalog/c4).
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2
 
3
- We prepared five variants of the data: `en`, `en.noclean`, `en.noblocklist`, `realnewslike`, and `multilingual`.
 
 
 
 
 
 
 
 
 
 
 
 
4
 
5
  For reference, these are the sizes of the variants:
6
 
@@ -8,13 +1123,76 @@ For reference, these are the sizes of the variants:
8
  - `en.noclean`: 2.3TB
9
  - `en.noblocklist`: 380GB
10
  - `realnewslike`: 15GB
11
- - `multilingual`: 9.7TB
12
 
13
  The `en.noblocklist` variant is exactly the same as the `en` variant, except we turned off the so-called "badwords filter", which removes all documents that contain words from the lists at https://github.com/LDNOOBW/List-of-Dirty-Naughty-Obscene-and-Otherwise-Bad-Words.
14
 
15
- # How do I download this?
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
16
 
17
- Unfortunately we ran out of time making this into a proper Huggingface dataset, accessible through the `datasets` Python package. Until we get that ready, please use git to do the download. First, make sure you have [Git Large File Storage](https://git-lfs.github.com) installed. Once that is done, downloading the whole dataset, all three variants, is easy:
18
 
19
  ```bash
20
  git clone https://huggingface.co/datasets/allenai/c4
@@ -34,10 +1212,179 @@ The `git clone` command in this variant will download a bunch of stub files that
34
  git lfs pull --include "multilingual/c4-nl.*.json.gz"
35
  ```
36
 
37
- # Acknowledgements
38
 
39
- Big ups to the good folks at [Common Crawl](https://commoncrawl.org) whose data made this possible ([consider donating](http://commoncrawl.org/donate/)!), to Google for creating the code that curates and filters the data, and to Huggingface, who had no issue with hosting these 3TB of data for public download!
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
40
 
41
- ### License
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
42
 
43
  We are releasing this dataset under the terms of [ODC-BY](https://opendatacommons.org/licenses/by/1-0/). By using this, you are also bound by the [Common Crawl terms of use](https://commoncrawl.org/terms-of-use/) in respect of the content contained in the dataset.
 
 
 
 
 
1
+ ---
2
+ pretty_name: C4
3
+ annotations_creators:
4
+ - no-annotation
5
+ language_creators:
6
+ - found
7
+ language:
8
+ - af
9
+ - am
10
+ - ar
11
+ - az
12
+ - be
13
+ - bg
14
+ - bn
15
+ - ca
16
+ - ceb
17
+ - co
18
+ - cs
19
+ - cy
20
+ - da
21
+ - de
22
+ - el
23
+ - en
24
+ - eo
25
+ - es
26
+ - et
27
+ - eu
28
+ - fa
29
+ - fi
30
+ - fil
31
+ - fr
32
+ - fy
33
+ - ga
34
+ - gd
35
+ - gl
36
+ - gu
37
+ - ha
38
+ - haw
39
+ - he
40
+ - hi
41
+ - hmn
42
+ - ht
43
+ - hu
44
+ - hy
45
+ - id
46
+ - ig
47
+ - is
48
+ - it
49
+ - iw
50
+ - ja
51
+ - jv
52
+ - ka
53
+ - kk
54
+ - km
55
+ - kn
56
+ - ko
57
+ - ku
58
+ - ky
59
+ - la
60
+ - lb
61
+ - lo
62
+ - lt
63
+ - lv
64
+ - mg
65
+ - mi
66
+ - mk
67
+ - ml
68
+ - mn
69
+ - mr
70
+ - ms
71
+ - mt
72
+ - my
73
+ - ne
74
+ - nl
75
+ - 'no'
76
+ - ny
77
+ - pa
78
+ - pl
79
+ - ps
80
+ - pt
81
+ - ro
82
+ - ru
83
+ - sd
84
+ - si
85
+ - sk
86
+ - sl
87
+ - sm
88
+ - sn
89
+ - so
90
+ - sq
91
+ - sr
92
+ - st
93
+ - su
94
+ - sv
95
+ - sw
96
+ - ta
97
+ - te
98
+ - tg
99
+ - th
100
+ - tr
101
+ - uk
102
+ - und
103
+ - ur
104
+ - uz
105
+ - vi
106
+ - xh
107
+ - yi
108
+ - yo
109
+ - zh
110
+ - zu
111
+ language_bcp47:
112
+ - bg-Latn
113
+ - el-Latn
114
+ - hi-Latn
115
+ - ja-Latn
116
+ - ru-Latn
117
+ - zh-Latn
118
+ license:
119
+ - odc-by
120
+ multilinguality:
121
+ - multilingual
122
+ size_categories:
123
+ - n<1K
124
+ - 1K<n<10K
125
+ - 10K<n<100K
126
+ - 100K<n<1M
127
+ - 1M<n<10M
128
+ - 10M<n<100M
129
+ - 100M<n<1B
130
+ - 1B<n<10B
131
+ source_datasets:
132
+ - original
133
+ task_categories:
134
+ - text-generation
135
+ - fill-mask
136
+ task_ids:
137
+ - language-modeling
138
+ - masked-language-modeling
139
+ paperswithcode_id: c4
140
+ dataset_info:
141
+ - config_name: en
142
+ features:
143
+ - name: text
144
+ dtype: string
145
+ - name: timestamp
146
+ dtype: string
147
+ - name: url
148
+ dtype: string
149
+ splits:
150
+ - name: train
151
+ num_bytes: 828589180707
152
+ num_examples: 364868892
153
+ - name: validation
154
+ num_bytes: 825767266
155
+ num_examples: 364608
156
+ download_size: 326778635540
157
+ dataset_size: 1657178361414
158
+ - config_name: en.noblocklist
159
+ features:
160
+ - name: text
161
+ dtype: string
162
+ - name: timestamp
163
+ dtype: string
164
+ - name: url
165
+ dtype: string
166
+ splits:
167
+ - name: train
168
+ num_bytes: 1029628201361
169
+ num_examples: 393391519
170
+ - name: validation
171
+ num_bytes: 1025606012
172
+ num_examples: 393226
173
+ download_size: 406611392434
174
+ dataset_size: 2059256402722
175
+ - config_name: realnewslike
176
+ features:
177
+ - name: text
178
+ dtype: string
179
+ - name: timestamp
180
+ dtype: string
181
+ - name: url
182
+ dtype: string
183
+ splits:
184
+ - name: train
185
+ num_bytes: 38165657946
186
+ num_examples: 13799838
187
+ - name: validation
188
+ num_bytes: 37875873
189
+ num_examples: 13863
190
+ download_size: 15419740744
191
+ dataset_size: 76331315892
192
+ - config_name: en.noclean
193
+ features:
194
+ - name: text
195
+ dtype: string
196
+ - name: timestamp
197
+ dtype: string
198
+ - name: url
199
+ dtype: string
200
+ splits:
201
+ - name: train
202
+ num_bytes: 6715509699938
203
+ num_examples: 1063805381
204
+ - name: validation
205
+ num_bytes: 6706356913
206
+ num_examples: 1065029
207
+ download_size: 2430376268625
208
+ dataset_size: 6722216056851
209
+ configs:
210
+ - config_name: en
211
+ data_files:
212
+ - split: train
213
+ path: en/c4-train.*.json.gz
214
+ - split: validation
215
+ path: en/c4-validation.*.json.gz
216
+ - config_name: en.noblocklist
217
+ data_files:
218
+ - split: train
219
+ path: en.noblocklist/c4-train.*.json.gz
220
+ - split: validation
221
+ path: en.noblocklist/c4-validation.*.json.gz
222
+ - config_name: en.noclean
223
+ data_files:
224
+ - split: train
225
+ path: en.noclean/c4-train.*.json.gz
226
+ - split: validation
227
+ path: en.noclean/c4-validation.*.json.gz
228
+ - config_name: realnewslike
229
+ data_files:
230
+ - split: train
231
+ path: realnewslike/c4-train.*.json.gz
232
+ - split: validation
233
+ path: realnewslike/c4-validation.*.json.gz
234
+ - config_name: multilingual
235
+ data_files:
236
+ - split: train
237
+ path:
238
+ - multilingual/c4-af.*.json.gz
239
+ - multilingual/c4-am.*.json.gz
240
+ - multilingual/c4-ar.*.json.gz
241
+ - multilingual/c4-az.*.json.gz
242
+ - multilingual/c4-be.*.json.gz
243
+ - multilingual/c4-bg.*.json.gz
244
+ - multilingual/c4-bg-Latn.*.json.gz
245
+ - multilingual/c4-bn.*.json.gz
246
+ - multilingual/c4-ca.*.json.gz
247
+ - multilingual/c4-ceb.*.json.gz
248
+ - multilingual/c4-co.*.json.gz
249
+ - multilingual/c4-cs.*.json.gz
250
+ - multilingual/c4-cy.*.json.gz
251
+ - multilingual/c4-da.*.json.gz
252
+ - multilingual/c4-de.*.json.gz
253
+ - multilingual/c4-el.*.json.gz
254
+ - multilingual/c4-el-Latn.*.json.gz
255
+ - multilingual/c4-en.*.json.gz
256
+ - multilingual/c4-eo.*.json.gz
257
+ - multilingual/c4-es.*.json.gz
258
+ - multilingual/c4-et.*.json.gz
259
+ - multilingual/c4-eu.*.json.gz
260
+ - multilingual/c4-fa.*.json.gz
261
+ - multilingual/c4-fi.*.json.gz
262
+ - multilingual/c4-fil.*.json.gz
263
+ - multilingual/c4-fr.*.json.gz
264
+ - multilingual/c4-fy.*.json.gz
265
+ - multilingual/c4-ga.*.json.gz
266
+ - multilingual/c4-gd.*.json.gz
267
+ - multilingual/c4-gl.*.json.gz
268
+ - multilingual/c4-gu.*.json.gz
269
+ - multilingual/c4-ha.*.json.gz
270
+ - multilingual/c4-haw.*.json.gz
271
+ - multilingual/c4-hi.*.json.gz
272
+ - multilingual/c4-hi-Latn.*.json.gz
273
+ - multilingual/c4-hmn.*.json.gz
274
+ - multilingual/c4-ht.*.json.gz
275
+ - multilingual/c4-hu.*.json.gz
276
+ - multilingual/c4-hy.*.json.gz
277
+ - multilingual/c4-id.*.json.gz
278
+ - multilingual/c4-ig.*.json.gz
279
+ - multilingual/c4-is.*.json.gz
280
+ - multilingual/c4-it.*.json.gz
281
+ - multilingual/c4-iw.*.json.gz
282
+ - multilingual/c4-ja.*.json.gz
283
+ - multilingual/c4-ja-Latn.*.json.gz
284
+ - multilingual/c4-jv.*.json.gz
285
+ - multilingual/c4-ka.*.json.gz
286
+ - multilingual/c4-kk.*.json.gz
287
+ - multilingual/c4-km.*.json.gz
288
+ - multilingual/c4-kn.*.json.gz
289
+ - multilingual/c4-ko.*.json.gz
290
+ - multilingual/c4-ku.*.json.gz
291
+ - multilingual/c4-ky.*.json.gz
292
+ - multilingual/c4-la.*.json.gz
293
+ - multilingual/c4-lb.*.json.gz
294
+ - multilingual/c4-lo.*.json.gz
295
+ - multilingual/c4-lt.*.json.gz
296
+ - multilingual/c4-lv.*.json.gz
297
+ - multilingual/c4-mg.*.json.gz
298
+ - multilingual/c4-mi.*.json.gz
299
+ - multilingual/c4-mk.*.json.gz
300
+ - multilingual/c4-ml.*.json.gz
301
+ - multilingual/c4-mn.*.json.gz
302
+ - multilingual/c4-mr.*.json.gz
303
+ - multilingual/c4-ms.*.json.gz
304
+ - multilingual/c4-mt.*.json.gz
305
+ - multilingual/c4-my.*.json.gz
306
+ - multilingual/c4-ne.*.json.gz
307
+ - multilingual/c4-nl.*.json.gz
308
+ - multilingual/c4-no.*.json.gz
309
+ - multilingual/c4-ny.*.json.gz
310
+ - multilingual/c4-pa.*.json.gz
311
+ - multilingual/c4-pl.*.json.gz
312
+ - multilingual/c4-ps.*.json.gz
313
+ - multilingual/c4-pt.*.json.gz
314
+ - multilingual/c4-ro.*.json.gz
315
+ - multilingual/c4-ru.*.json.gz
316
+ - multilingual/c4-ru-Latn.*.json.gz
317
+ - multilingual/c4-sd.*.json.gz
318
+ - multilingual/c4-si.*.json.gz
319
+ - multilingual/c4-sk.*.json.gz
320
+ - multilingual/c4-sl.*.json.gz
321
+ - multilingual/c4-sm.*.json.gz
322
+ - multilingual/c4-sn.*.json.gz
323
+ - multilingual/c4-so.*.json.gz
324
+ - multilingual/c4-sq.*.json.gz
325
+ - multilingual/c4-sr.*.json.gz
326
+ - multilingual/c4-st.*.json.gz
327
+ - multilingual/c4-su.*.json.gz
328
+ - multilingual/c4-sv.*.json.gz
329
+ - multilingual/c4-sw.*.json.gz
330
+ - multilingual/c4-ta.*.json.gz
331
+ - multilingual/c4-te.*.json.gz
332
+ - multilingual/c4-tg.*.json.gz
333
+ - multilingual/c4-th.*.json.gz
334
+ - multilingual/c4-tr.*.json.gz
335
+ - multilingual/c4-uk.*.json.gz
336
+ - multilingual/c4-und.*.json.gz
337
+ - multilingual/c4-ur.*.json.gz
338
+ - multilingual/c4-uz.*.json.gz
339
+ - multilingual/c4-vi.*.json.gz
340
+ - multilingual/c4-xh.*.json.gz
341
+ - multilingual/c4-yi.*.json.gz
342
+ - multilingual/c4-yo.*.json.gz
343
+ - multilingual/c4-zh.*.json.gz
344
+ - multilingual/c4-zh-Latn.*.json.gz
345
+ - multilingual/c4-zu.*.json.gz
346
+ - split: validation
347
+ path:
348
+ - multilingual/c4-af-validation.*.json.gz
349
+ - multilingual/c4-am-validation.*.json.gz
350
+ - multilingual/c4-ar-validation.*.json.gz
351
+ - multilingual/c4-az-validation.*.json.gz
352
+ - multilingual/c4-be-validation.*.json.gz
353
+ - multilingual/c4-bg-validation.*.json.gz
354
+ - multilingual/c4-bg-Latn-validation.*.json.gz
355
+ - multilingual/c4-bn-validation.*.json.gz
356
+ - multilingual/c4-ca-validation.*.json.gz
357
+ - multilingual/c4-ceb-validation.*.json.gz
358
+ - multilingual/c4-co-validation.*.json.gz
359
+ - multilingual/c4-cs-validation.*.json.gz
360
+ - multilingual/c4-cy-validation.*.json.gz
361
+ - multilingual/c4-da-validation.*.json.gz
362
+ - multilingual/c4-de-validation.*.json.gz
363
+ - multilingual/c4-el-validation.*.json.gz
364
+ - multilingual/c4-el-Latn-validation.*.json.gz
365
+ - multilingual/c4-en-validation.*.json.gz
366
+ - multilingual/c4-eo-validation.*.json.gz
367
+ - multilingual/c4-es-validation.*.json.gz
368
+ - multilingual/c4-et-validation.*.json.gz
369
+ - multilingual/c4-eu-validation.*.json.gz
370
+ - multilingual/c4-fa-validation.*.json.gz
371
+ - multilingual/c4-fi-validation.*.json.gz
372
+ - multilingual/c4-fil-validation.*.json.gz
373
+ - multilingual/c4-fr-validation.*.json.gz
374
+ - multilingual/c4-fy-validation.*.json.gz
375
+ - multilingual/c4-ga-validation.*.json.gz
376
+ - multilingual/c4-gd-validation.*.json.gz
377
+ - multilingual/c4-gl-validation.*.json.gz
378
+ - multilingual/c4-gu-validation.*.json.gz
379
+ - multilingual/c4-ha-validation.*.json.gz
380
+ - multilingual/c4-haw-validation.*.json.gz
381
+ - multilingual/c4-hi-validation.*.json.gz
382
+ - multilingual/c4-hi-Latn-validation.*.json.gz
383
+ - multilingual/c4-hmn-validation.*.json.gz
384
+ - multilingual/c4-ht-validation.*.json.gz
385
+ - multilingual/c4-hu-validation.*.json.gz
386
+ - multilingual/c4-hy-validation.*.json.gz
387
+ - multilingual/c4-id-validation.*.json.gz
388
+ - multilingual/c4-ig-validation.*.json.gz
389
+ - multilingual/c4-is-validation.*.json.gz
390
+ - multilingual/c4-it-validation.*.json.gz
391
+ - multilingual/c4-iw-validation.*.json.gz
392
+ - multilingual/c4-ja-validation.*.json.gz
393
+ - multilingual/c4-ja-Latn-validation.*.json.gz
394
+ - multilingual/c4-jv-validation.*.json.gz
395
+ - multilingual/c4-ka-validation.*.json.gz
396
+ - multilingual/c4-kk-validation.*.json.gz
397
+ - multilingual/c4-km-validation.*.json.gz
398
+ - multilingual/c4-kn-validation.*.json.gz
399
+ - multilingual/c4-ko-validation.*.json.gz
400
+ - multilingual/c4-ku-validation.*.json.gz
401
+ - multilingual/c4-ky-validation.*.json.gz
402
+ - multilingual/c4-la-validation.*.json.gz
403
+ - multilingual/c4-lb-validation.*.json.gz
404
+ - multilingual/c4-lo-validation.*.json.gz
405
+ - multilingual/c4-lt-validation.*.json.gz
406
+ - multilingual/c4-lv-validation.*.json.gz
407
+ - multilingual/c4-mg-validation.*.json.gz
408
+ - multilingual/c4-mi-validation.*.json.gz
409
+ - multilingual/c4-mk-validation.*.json.gz
410
+ - multilingual/c4-ml-validation.*.json.gz
411
+ - multilingual/c4-mn-validation.*.json.gz
412
+ - multilingual/c4-mr-validation.*.json.gz
413
+ - multilingual/c4-ms-validation.*.json.gz
414
+ - multilingual/c4-mt-validation.*.json.gz
415
+ - multilingual/c4-my-validation.*.json.gz
416
+ - multilingual/c4-ne-validation.*.json.gz
417
+ - multilingual/c4-nl-validation.*.json.gz
418
+ - multilingual/c4-no-validation.*.json.gz
419
+ - multilingual/c4-ny-validation.*.json.gz
420
+ - multilingual/c4-pa-validation.*.json.gz
421
+ - multilingual/c4-pl-validation.*.json.gz
422
+ - multilingual/c4-ps-validation.*.json.gz
423
+ - multilingual/c4-pt-validation.*.json.gz
424
+ - multilingual/c4-ro-validation.*.json.gz
425
+ - multilingual/c4-ru-validation.*.json.gz
426
+ - multilingual/c4-ru-Latn-validation.*.json.gz
427
+ - multilingual/c4-sd-validation.*.json.gz
428
+ - multilingual/c4-si-validation.*.json.gz
429
+ - multilingual/c4-sk-validation.*.json.gz
430
+ - multilingual/c4-sl-validation.*.json.gz
431
+ - multilingual/c4-sm-validation.*.json.gz
432
+ - multilingual/c4-sn-validation.*.json.gz
433
+ - multilingual/c4-so-validation.*.json.gz
434
+ - multilingual/c4-sq-validation.*.json.gz
435
+ - multilingual/c4-sr-validation.*.json.gz
436
+ - multilingual/c4-st-validation.*.json.gz
437
+ - multilingual/c4-su-validation.*.json.gz
438
+ - multilingual/c4-sv-validation.*.json.gz
439
+ - multilingual/c4-sw-validation.*.json.gz
440
+ - multilingual/c4-ta-validation.*.json.gz
441
+ - multilingual/c4-te-validation.*.json.gz
442
+ - multilingual/c4-tg-validation.*.json.gz
443
+ - multilingual/c4-th-validation.*.json.gz
444
+ - multilingual/c4-tr-validation.*.json.gz
445
+ - multilingual/c4-uk-validation.*.json.gz
446
+ - multilingual/c4-und-validation.*.json.gz
447
+ - multilingual/c4-ur-validation.*.json.gz
448
+ - multilingual/c4-uz-validation.*.json.gz
449
+ - multilingual/c4-vi-validation.*.json.gz
450
+ - multilingual/c4-xh-validation.*.json.gz
451
+ - multilingual/c4-yi-validation.*.json.gz
452
+ - multilingual/c4-yo-validation.*.json.gz
453
+ - multilingual/c4-zh-validation.*.json.gz
454
+ - multilingual/c4-zh-Latn-validation.*.json.gz
455
+ - multilingual/c4-zu-validation.*.json.gz
456
+ - config_name: af
457
+ data_files:
458
+ - split: train
459
+ path: multilingual/c4-af.*.json.gz
460
+ - split: validation
461
+ path: multilingual/c4-af-validation.*.json.gz
462
+ - config_name: am
463
+ data_files:
464
+ - split: train
465
+ path: multilingual/c4-am.*.json.gz
466
+ - split: validation
467
+ path: multilingual/c4-am-validation.*.json.gz
468
+ - config_name: ar
469
+ data_files:
470
+ - split: train
471
+ path: multilingual/c4-ar.*.json.gz
472
+ - split: validation
473
+ path: multilingual/c4-ar-validation.*.json.gz
474
+ - config_name: az
475
+ data_files:
476
+ - split: train
477
+ path: multilingual/c4-az.*.json.gz
478
+ - split: validation
479
+ path: multilingual/c4-az-validation.*.json.gz
480
+ - config_name: be
481
+ data_files:
482
+ - split: train
483
+ path: multilingual/c4-be.*.json.gz
484
+ - split: validation
485
+ path: multilingual/c4-be-validation.*.json.gz
486
+ - config_name: bg
487
+ data_files:
488
+ - split: train
489
+ path: multilingual/c4-bg.*.json.gz
490
+ - split: validation
491
+ path: multilingual/c4-bg-validation.*.json.gz
492
+ - config_name: bg-Latn
493
+ data_files:
494
+ - split: train
495
+ path: multilingual/c4-bg-Latn.*.json.gz
496
+ - split: validation
497
+ path: multilingual/c4-bg-Latn-validation.*.json.gz
498
+ - config_name: bn
499
+ data_files:
500
+ - split: train
501
+ path: multilingual/c4-bn.*.json.gz
502
+ - split: validation
503
+ path: multilingual/c4-bn-validation.*.json.gz
504
+ - config_name: ca
505
+ data_files:
506
+ - split: train
507
+ path: multilingual/c4-ca.*.json.gz
508
+ - split: validation
509
+ path: multilingual/c4-ca-validation.*.json.gz
510
+ - config_name: ceb
511
+ data_files:
512
+ - split: train
513
+ path: multilingual/c4-ceb.*.json.gz
514
+ - split: validation
515
+ path: multilingual/c4-ceb-validation.*.json.gz
516
+ - config_name: co
517
+ data_files:
518
+ - split: train
519
+ path: multilingual/c4-co.*.json.gz
520
+ - split: validation
521
+ path: multilingual/c4-co-validation.*.json.gz
522
+ - config_name: cs
523
+ data_files:
524
+ - split: train
525
+ path: multilingual/c4-cs.*.json.gz
526
+ - split: validation
527
+ path: multilingual/c4-cs-validation.*.json.gz
528
+ - config_name: cy
529
+ data_files:
530
+ - split: train
531
+ path: multilingual/c4-cy.*.json.gz
532
+ - split: validation
533
+ path: multilingual/c4-cy-validation.*.json.gz
534
+ - config_name: da
535
+ data_files:
536
+ - split: train
537
+ path: multilingual/c4-da.*.json.gz
538
+ - split: validation
539
+ path: multilingual/c4-da-validation.*.json.gz
540
+ - config_name: de
541
+ data_files:
542
+ - split: train
543
+ path: multilingual/c4-de.*.json.gz
544
+ - split: validation
545
+ path: multilingual/c4-de-validation.*.json.gz
546
+ - config_name: el
547
+ data_files:
548
+ - split: train
549
+ path: multilingual/c4-el.*.json.gz
550
+ - split: validation
551
+ path: multilingual/c4-el-validation.*.json.gz
552
+ - config_name: el-Latn
553
+ data_files:
554
+ - split: train
555
+ path: multilingual/c4-el-Latn.*.json.gz
556
+ - split: validation
557
+ path: multilingual/c4-el-Latn-validation.*.json.gz
558
+ - config_name: en-multi
559
+ data_files:
560
+ - split: train
561
+ path: multilingual/c4-en.*.json.gz
562
+ - split: validation
563
+ path: multilingual/c4-en-validation.*.json.gz
564
+ - config_name: eo
565
+ data_files:
566
+ - split: train
567
+ path: multilingual/c4-eo.*.json.gz
568
+ - split: validation
569
+ path: multilingual/c4-eo-validation.*.json.gz
570
+ - config_name: es
571
+ data_files:
572
+ - split: train
573
+ path: multilingual/c4-es.*.json.gz
574
+ - split: validation
575
+ path: multilingual/c4-es-validation.*.json.gz
576
+ - config_name: et
577
+ data_files:
578
+ - split: train
579
+ path: multilingual/c4-et.*.json.gz
580
+ - split: validation
581
+ path: multilingual/c4-et-validation.*.json.gz
582
+ - config_name: eu
583
+ data_files:
584
+ - split: train
585
+ path: multilingual/c4-eu.*.json.gz
586
+ - split: validation
587
+ path: multilingual/c4-eu-validation.*.json.gz
588
+ - config_name: fa
589
+ data_files:
590
+ - split: train
591
+ path: multilingual/c4-fa.*.json.gz
592
+ - split: validation
593
+ path: multilingual/c4-fa-validation.*.json.gz
594
+ - config_name: fi
595
+ data_files:
596
+ - split: train
597
+ path: multilingual/c4-fi.*.json.gz
598
+ - split: validation
599
+ path: multilingual/c4-fi-validation.*.json.gz
600
+ - config_name: fil
601
+ data_files:
602
+ - split: train
603
+ path: multilingual/c4-fil.*.json.gz
604
+ - split: validation
605
+ path: multilingual/c4-fil-validation.*.json.gz
606
+ - config_name: fr
607
+ data_files:
608
+ - split: train
609
+ path: multilingual/c4-fr.*.json.gz
610
+ - split: validation
611
+ path: multilingual/c4-fr-validation.*.json.gz
612
+ - config_name: fy
613
+ data_files:
614
+ - split: train
615
+ path: multilingual/c4-fy.*.json.gz
616
+ - split: validation
617
+ path: multilingual/c4-fy-validation.*.json.gz
618
+ - config_name: ga
619
+ data_files:
620
+ - split: train
621
+ path: multilingual/c4-ga.*.json.gz
622
+ - split: validation
623
+ path: multilingual/c4-ga-validation.*.json.gz
624
+ - config_name: gd
625
+ data_files:
626
+ - split: train
627
+ path: multilingual/c4-gd.*.json.gz
628
+ - split: validation
629
+ path: multilingual/c4-gd-validation.*.json.gz
630
+ - config_name: gl
631
+ data_files:
632
+ - split: train
633
+ path: multilingual/c4-gl.*.json.gz
634
+ - split: validation
635
+ path: multilingual/c4-gl-validation.*.json.gz
636
+ - config_name: gu
637
+ data_files:
638
+ - split: train
639
+ path: multilingual/c4-gu.*.json.gz
640
+ - split: validation
641
+ path: multilingual/c4-gu-validation.*.json.gz
642
+ - config_name: ha
643
+ data_files:
644
+ - split: train
645
+ path: multilingual/c4-ha.*.json.gz
646
+ - split: validation
647
+ path: multilingual/c4-ha-validation.*.json.gz
648
+ - config_name: haw
649
+ data_files:
650
+ - split: train
651
+ path: multilingual/c4-haw.*.json.gz
652
+ - split: validation
653
+ path: multilingual/c4-haw-validation.*.json.gz
654
+ - config_name: hi
655
+ data_files:
656
+ - split: train
657
+ path: multilingual/c4-hi.*.json.gz
658
+ - split: validation
659
+ path: multilingual/c4-hi-validation.*.json.gz
660
+ - config_name: hi-Latn
661
+ data_files:
662
+ - split: train
663
+ path: multilingual/c4-hi-Latn.*.json.gz
664
+ - split: validation
665
+ path: multilingual/c4-hi-Latn-validation.*.json.gz
666
+ - config_name: hmn
667
+ data_files:
668
+ - split: train
669
+ path: multilingual/c4-hmn.*.json.gz
670
+ - split: validation
671
+ path: multilingual/c4-hmn-validation.*.json.gz
672
+ - config_name: ht
673
+ data_files:
674
+ - split: train
675
+ path: multilingual/c4-ht.*.json.gz
676
+ - split: validation
677
+ path: multilingual/c4-ht-validation.*.json.gz
678
+ - config_name: hu
679
+ data_files:
680
+ - split: train
681
+ path: multilingual/c4-hu.*.json.gz
682
+ - split: validation
683
+ path: multilingual/c4-hu-validation.*.json.gz
684
+ - config_name: hy
685
+ data_files:
686
+ - split: train
687
+ path: multilingual/c4-hy.*.json.gz
688
+ - split: validation
689
+ path: multilingual/c4-hy-validation.*.json.gz
690
+ - config_name: id
691
+ data_files:
692
+ - split: train
693
+ path: multilingual/c4-id.*.json.gz
694
+ - split: validation
695
+ path: multilingual/c4-id-validation.*.json.gz
696
+ - config_name: ig
697
+ data_files:
698
+ - split: train
699
+ path: multilingual/c4-ig.*.json.gz
700
+ - split: validation
701
+ path: multilingual/c4-ig-validation.*.json.gz
702
+ - config_name: is
703
+ data_files:
704
+ - split: train
705
+ path: multilingual/c4-is.*.json.gz
706
+ - split: validation
707
+ path: multilingual/c4-is-validation.*.json.gz
708
+ - config_name: it
709
+ data_files:
710
+ - split: train
711
+ path: multilingual/c4-it.*.json.gz
712
+ - split: validation
713
+ path: multilingual/c4-it-validation.*.json.gz
714
+ - config_name: iw
715
+ data_files:
716
+ - split: train
717
+ path: multilingual/c4-iw.*.json.gz
718
+ - split: validation
719
+ path: multilingual/c4-iw-validation.*.json.gz
720
+ - config_name: ja
721
+ data_files:
722
+ - split: train
723
+ path: multilingual/c4-ja.*.json.gz
724
+ - split: validation
725
+ path: multilingual/c4-ja-validation.*.json.gz
726
+ - config_name: ja-Latn
727
+ data_files:
728
+ - split: train
729
+ path: multilingual/c4-ja-Latn.*.json.gz
730
+ - split: validation
731
+ path: multilingual/c4-ja-Latn-validation.*.json.gz
732
+ - config_name: jv
733
+ data_files:
734
+ - split: train
735
+ path: multilingual/c4-jv.*.json.gz
736
+ - split: validation
737
+ path: multilingual/c4-jv-validation.*.json.gz
738
+ - config_name: ka
739
+ data_files:
740
+ - split: train
741
+ path: multilingual/c4-ka.*.json.gz
742
+ - split: validation
743
+ path: multilingual/c4-ka-validation.*.json.gz
744
+ - config_name: kk
745
+ data_files:
746
+ - split: train
747
+ path: multilingual/c4-kk.*.json.gz
748
+ - split: validation
749
+ path: multilingual/c4-kk-validation.*.json.gz
750
+ - config_name: km
751
+ data_files:
752
+ - split: train
753
+ path: multilingual/c4-km.*.json.gz
754
+ - split: validation
755
+ path: multilingual/c4-km-validation.*.json.gz
756
+ - config_name: kn
757
+ data_files:
758
+ - split: train
759
+ path: multilingual/c4-kn.*.json.gz
760
+ - split: validation
761
+ path: multilingual/c4-kn-validation.*.json.gz
762
+ - config_name: ko
763
+ data_files:
764
+ - split: train
765
+ path: multilingual/c4-ko.*.json.gz
766
+ - split: validation
767
+ path: multilingual/c4-ko-validation.*.json.gz
768
+ - config_name: ku
769
+ data_files:
770
+ - split: train
771
+ path: multilingual/c4-ku.*.json.gz
772
+ - split: validation
773
+ path: multilingual/c4-ku-validation.*.json.gz
774
+ - config_name: ky
775
+ data_files:
776
+ - split: train
777
+ path: multilingual/c4-ky.*.json.gz
778
+ - split: validation
779
+ path: multilingual/c4-ky-validation.*.json.gz
780
+ - config_name: la
781
+ data_files:
782
+ - split: train
783
+ path: multilingual/c4-la.*.json.gz
784
+ - split: validation
785
+ path: multilingual/c4-la-validation.*.json.gz
786
+ - config_name: lb
787
+ data_files:
788
+ - split: train
789
+ path: multilingual/c4-lb.*.json.gz
790
+ - split: validation
791
+ path: multilingual/c4-lb-validation.*.json.gz
792
+ - config_name: lo
793
+ data_files:
794
+ - split: train
795
+ path: multilingual/c4-lo.*.json.gz
796
+ - split: validation
797
+ path: multilingual/c4-lo-validation.*.json.gz
798
+ - config_name: lt
799
+ data_files:
800
+ - split: train
801
+ path: multilingual/c4-lt.*.json.gz
802
+ - split: validation
803
+ path: multilingual/c4-lt-validation.*.json.gz
804
+ - config_name: lv
805
+ data_files:
806
+ - split: train
807
+ path: multilingual/c4-lv.*.json.gz
808
+ - split: validation
809
+ path: multilingual/c4-lv-validation.*.json.gz
810
+ - config_name: mg
811
+ data_files:
812
+ - split: train
813
+ path: multilingual/c4-mg.*.json.gz
814
+ - split: validation
815
+ path: multilingual/c4-mg-validation.*.json.gz
816
+ - config_name: mi
817
+ data_files:
818
+ - split: train
819
+ path: multilingual/c4-mi.*.json.gz
820
+ - split: validation
821
+ path: multilingual/c4-mi-validation.*.json.gz
822
+ - config_name: mk
823
+ data_files:
824
+ - split: train
825
+ path: multilingual/c4-mk.*.json.gz
826
+ - split: validation
827
+ path: multilingual/c4-mk-validation.*.json.gz
828
+ - config_name: ml
829
+ data_files:
830
+ - split: train
831
+ path: multilingual/c4-ml.*.json.gz
832
+ - split: validation
833
+ path: multilingual/c4-ml-validation.*.json.gz
834
+ - config_name: mn
835
+ data_files:
836
+ - split: train
837
+ path: multilingual/c4-mn.*.json.gz
838
+ - split: validation
839
+ path: multilingual/c4-mn-validation.*.json.gz
840
+ - config_name: mr
841
+ data_files:
842
+ - split: train
843
+ path: multilingual/c4-mr.*.json.gz
844
+ - split: validation
845
+ path: multilingual/c4-mr-validation.*.json.gz
846
+ - config_name: ms
847
+ data_files:
848
+ - split: train
849
+ path: multilingual/c4-ms.*.json.gz
850
+ - split: validation
851
+ path: multilingual/c4-ms-validation.*.json.gz
852
+ - config_name: mt
853
+ data_files:
854
+ - split: train
855
+ path: multilingual/c4-mt.*.json.gz
856
+ - split: validation
857
+ path: multilingual/c4-mt-validation.*.json.gz
858
+ - config_name: my
859
+ data_files:
860
+ - split: train
861
+ path: multilingual/c4-my.*.json.gz
862
+ - split: validation
863
+ path: multilingual/c4-my-validation.*.json.gz
864
+ - config_name: ne
865
+ data_files:
866
+ - split: train
867
+ path: multilingual/c4-ne.*.json.gz
868
+ - split: validation
869
+ path: multilingual/c4-ne-validation.*.json.gz
870
+ - config_name: nl
871
+ data_files:
872
+ - split: train
873
+ path: multilingual/c4-nl.*.json.gz
874
+ - split: validation
875
+ path: multilingual/c4-nl-validation.*.json.gz
876
+ - config_name: 'no'
877
+ data_files:
878
+ - split: train
879
+ path: multilingual/c4-no.*.json.gz
880
+ - split: validation
881
+ path: multilingual/c4-no-validation.*.json.gz
882
+ - config_name: ny
883
+ data_files:
884
+ - split: train
885
+ path: multilingual/c4-ny.*.json.gz
886
+ - split: validation
887
+ path: multilingual/c4-ny-validation.*.json.gz
888
+ - config_name: pa
889
+ data_files:
890
+ - split: train
891
+ path: multilingual/c4-pa.*.json.gz
892
+ - split: validation
893
+ path: multilingual/c4-pa-validation.*.json.gz
894
+ - config_name: pl
895
+ data_files:
896
+ - split: train
897
+ path: multilingual/c4-pl.*.json.gz
898
+ - split: validation
899
+ path: multilingual/c4-pl-validation.*.json.gz
900
+ - config_name: ps
901
+ data_files:
902
+ - split: train
903
+ path: multilingual/c4-ps.*.json.gz
904
+ - split: validation
905
+ path: multilingual/c4-ps-validation.*.json.gz
906
+ - config_name: pt
907
+ data_files:
908
+ - split: train
909
+ path: multilingual/c4-pt.*.json.gz
910
+ - split: validation
911
+ path: multilingual/c4-pt-validation.*.json.gz
912
+ - config_name: ro
913
+ data_files:
914
+ - split: train
915
+ path: multilingual/c4-ro.*.json.gz
916
+ - split: validation
917
+ path: multilingual/c4-ro-validation.*.json.gz
918
+ - config_name: ru
919
+ data_files:
920
+ - split: train
921
+ path: multilingual/c4-ru.*.json.gz
922
+ - split: validation
923
+ path: multilingual/c4-ru-validation.*.json.gz
924
+ - config_name: ru-Latn
925
+ data_files:
926
+ - split: train
927
+ path: multilingual/c4-ru-Latn.*.json.gz
928
+ - split: validation
929
+ path: multilingual/c4-ru-Latn-validation.*.json.gz
930
+ - config_name: sd
931
+ data_files:
932
+ - split: train
933
+ path: multilingual/c4-sd.*.json.gz
934
+ - split: validation
935
+ path: multilingual/c4-sd-validation.*.json.gz
936
+ - config_name: si
937
+ data_files:
938
+ - split: train
939
+ path: multilingual/c4-si.*.json.gz
940
+ - split: validation
941
+ path: multilingual/c4-si-validation.*.json.gz
942
+ - config_name: sk
943
+ data_files:
944
+ - split: train
945
+ path: multilingual/c4-sk.*.json.gz
946
+ - split: validation
947
+ path: multilingual/c4-sk-validation.*.json.gz
948
+ - config_name: sl
949
+ data_files:
950
+ - split: train
951
+ path: multilingual/c4-sl.*.json.gz
952
+ - split: validation
953
+ path: multilingual/c4-sl-validation.*.json.gz
954
+ - config_name: sm
955
+ data_files:
956
+ - split: train
957
+ path: multilingual/c4-sm.*.json.gz
958
+ - split: validation
959
+ path: multilingual/c4-sm-validation.*.json.gz
960
+ - config_name: sn
961
+ data_files:
962
+ - split: train
963
+ path: multilingual/c4-sn.*.json.gz
964
+ - split: validation
965
+ path: multilingual/c4-sn-validation.*.json.gz
966
+ - config_name: so
967
+ data_files:
968
+ - split: train
969
+ path: multilingual/c4-so.*.json.gz
970
+ - split: validation
971
+ path: multilingual/c4-so-validation.*.json.gz
972
+ - config_name: sq
973
+ data_files:
974
+ - split: train
975
+ path: multilingual/c4-sq.*.json.gz
976
+ - split: validation
977
+ path: multilingual/c4-sq-validation.*.json.gz
978
+ - config_name: sr
979
+ data_files:
980
+ - split: train
981
+ path: multilingual/c4-sr.*.json.gz
982
+ - split: validation
983
+ path: multilingual/c4-sr-validation.*.json.gz
984
+ - config_name: st
985
+ data_files:
986
+ - split: train
987
+ path: multilingual/c4-st.*.json.gz
988
+ - split: validation
989
+ path: multilingual/c4-st-validation.*.json.gz
990
+ - config_name: su
991
+ data_files:
992
+ - split: train
993
+ path: multilingual/c4-su.*.json.gz
994
+ - split: validation
995
+ path: multilingual/c4-su-validation.*.json.gz
996
+ - config_name: sv
997
+ data_files:
998
+ - split: train
999
+ path: multilingual/c4-sv.*.json.gz
1000
+ - split: validation
1001
+ path: multilingual/c4-sv-validation.*.json.gz
1002
+ - config_name: sw
1003
+ data_files:
1004
+ - split: train
1005
+ path: multilingual/c4-sw.*.json.gz
1006
+ - split: validation
1007
+ path: multilingual/c4-sw-validation.*.json.gz
1008
+ - config_name: ta
1009
+ data_files:
1010
+ - split: train
1011
+ path: multilingual/c4-ta.*.json.gz
1012
+ - split: validation
1013
+ path: multilingual/c4-ta-validation.*.json.gz
1014
+ - config_name: te
1015
+ data_files:
1016
+ - split: train
1017
+ path: multilingual/c4-te.*.json.gz
1018
+ - split: validation
1019
+ path: multilingual/c4-te-validation.*.json.gz
1020
+ - config_name: tg
1021
+ data_files:
1022
+ - split: train
1023
+ path: multilingual/c4-tg.*.json.gz
1024
+ - split: validation
1025
+ path: multilingual/c4-tg-validation.*.json.gz
1026
+ - config_name: th
1027
+ data_files:
1028
+ - split: train
1029
+ path: multilingual/c4-th.*.json.gz
1030
+ - split: validation
1031
+ path: multilingual/c4-th-validation.*.json.gz
1032
+ - config_name: tr
1033
+ data_files:
1034
+ - split: train
1035
+ path: multilingual/c4-tr.*.json.gz
1036
+ - split: validation
1037
+ path: multilingual/c4-tr-validation.*.json.gz
1038
+ - config_name: uk
1039
+ data_files:
1040
+ - split: train
1041
+ path: multilingual/c4-uk.*.json.gz
1042
+ - split: validation
1043
+ path: multilingual/c4-uk-validation.*.json.gz
1044
+ - config_name: und
1045
+ data_files:
1046
+ - split: train
1047
+ path: multilingual/c4-und.*.json.gz
1048
+ - split: validation
1049
+ path: multilingual/c4-und-validation.*.json.gz
1050
+ - config_name: ur
1051
+ data_files:
1052
+ - split: train
1053
+ path: multilingual/c4-ur.*.json.gz
1054
+ - split: validation
1055
+ path: multilingual/c4-ur-validation.*.json.gz
1056
+ - config_name: uz
1057
+ data_files:
1058
+ - split: train
1059
+ path: multilingual/c4-uz.*.json.gz
1060
+ - split: validation
1061
+ path: multilingual/c4-uz-validation.*.json.gz
1062
+ - config_name: vi
1063
+ data_files:
1064
+ - split: train
1065
+ path: multilingual/c4-vi.*.json.gz
1066
+ - split: validation
1067
+ path: multilingual/c4-vi-validation.*.json.gz
1068
+ - config_name: xh
1069
+ data_files:
1070
+ - split: train
1071
+ path: multilingual/c4-xh.*.json.gz
1072
+ - split: validation
1073
+ path: multilingual/c4-xh-validation.*.json.gz
1074
+ - config_name: yi
1075
+ data_files:
1076
+ - split: train
1077
+ path: multilingual/c4-yi.*.json.gz
1078
+ - split: validation
1079
+ path: multilingual/c4-yi-validation.*.json.gz
1080
+ - config_name: yo
1081
+ data_files:
1082
+ - split: train
1083
+ path: multilingual/c4-yo.*.json.gz
1084
+ - split: validation
1085
+ path: multilingual/c4-yo-validation.*.json.gz
1086
+ - config_name: zh
1087
+ data_files:
1088
+ - split: train
1089
+ path: multilingual/c4-zh.*.json.gz
1090
+ - split: validation
1091
+ path: multilingual/c4-zh-validation.*.json.gz
1092
+ - config_name: zh-Latn
1093
+ data_files:
1094
+ - split: train
1095
+ path: multilingual/c4-zh-Latn.*.json.gz
1096
+ - split: validation
1097
+ path: multilingual/c4-zh-Latn-validation.*.json.gz
1098
+ - config_name: zu
1099
+ data_files:
1100
+ - split: train
1101
+ path: multilingual/c4-zu.*.json.gz
1102
+ - split: validation
1103
+ path: multilingual/c4-zu-validation.*.json.gz
1104
+ ---
1105
 
1106
+ # C4
1107
+
1108
+ ## Dataset Description
1109
+
1110
+ - **Paper:** https://arxiv.org/abs/1910.10683
1111
+
1112
+ ### Dataset Summary
1113
+
1114
+ A colossal, cleaned version of Common Crawl's web crawl corpus. Based on Common Crawl dataset: "https://commoncrawl.org".
1115
+
1116
+ This is the processed version of [Google's C4 dataset](https://www.tensorflow.org/datasets/catalog/c4)
1117
+
1118
+ We prepared five variants of the data: `en`, `en.noclean`, `en.noblocklist`, `realnewslike`, and `multilingual` (mC4).
1119
 
1120
  For reference, these are the sizes of the variants:
1121
 
 
1123
  - `en.noclean`: 2.3TB
1124
  - `en.noblocklist`: 380GB
1125
  - `realnewslike`: 15GB
1126
+ - `multilingual` (mC4): 9.7TB (108 subsets, one per language)
1127
 
1128
  The `en.noblocklist` variant is exactly the same as the `en` variant, except we turned off the so-called "badwords filter", which removes all documents that contain words from the lists at https://github.com/LDNOOBW/List-of-Dirty-Naughty-Obscene-and-Otherwise-Bad-Words.
1129
 
1130
+ #### How do I download this?
1131
+
1132
+ ##### Using 🤗 Datasets
1133
+
1134
+ ```python
1135
+ from datasets import load_dataset
1136
+
1137
+ # English only
1138
+ en = load_dataset("allenai/c4", "en")
1139
+
1140
+ # Other variants in english
1141
+ en_noclean = load_dataset("allenai/c4", "en.noclean")
1142
+ en_noblocklist = load_dataset("allenai/c4", "en.noblocklist")
1143
+ realnewslike = load_dataset("allenai/c4", "realnewslike")
1144
+
1145
+ # Multilingual (108 languages)
1146
+ multilingual = load_dataset("allenai/c4", "multilingual")
1147
+
1148
+ # One specific language
1149
+ es = load_dataset("allenai/c4", "es")
1150
+ ```
1151
+
1152
+ Since this dataset is big, it is encouraged to load it in streaming mode using `streaming=True`, for example:
1153
+
1154
+ ```python
1155
+ en = load_dataset("allenai/c4", "en", streaming=True)
1156
+ ```
1157
+
1158
+ You can also load and mix multiple languages:
1159
+
1160
+ ```python
1161
+ from datasets import concatenate_datasets, interleave_datasets, load_dataset
1162
+
1163
+ es = load_dataset("allenai/c4", "es", streaming=True)
1164
+ fr = load_dataset("allenai/c4", "fr", streaming=True)
1165
+
1166
+ # Concatenate both datasets
1167
+ concatenated = concatenate_datasets([es, fr])
1168
+ # Or interleave them (alternates between one and the other)
1169
+ interleaved = interleave_datasets([es, fr])
1170
+ ```
1171
+
1172
+ ##### Using Dask
1173
+
1174
+ ```python
1175
+ import dask.dataframe as dd
1176
+
1177
+ df = dd.read_json("hf://datasets/allenai/c4/en/c4-train.*.json.gz")
1178
+
1179
+ # English only
1180
+ en_df = dd.read_json("hf://datasets/allenai/c4/en/c4-*.json.gz")
1181
+
1182
+ # Other variants in english
1183
+ en_noclean_df = dd.read_json("hf://datasets/allenai/c4/en/noclean/c4-*.json.gz")
1184
+ en_noblocklist_df = dd.read_json("hf://datasets/allenai/c4/en.noblocklist/c4-*.json.gz")
1185
+ realnewslike_df = dd.read_json("hf://datasets/allenai/c4/realnewslike/c4-*.json.gz")
1186
+
1187
+ # Multilingual (108 languages)
1188
+ multilingual_df = dd.read_json("hf://datasets/allenai/c4/multilingual/c4-*.json.gz")
1189
+
1190
+ # One specific language
1191
+ es_train_df = dd.read_json("hf://datasets/allenai/c4/multilingual/c4-es.*.json.gz")
1192
+ es_valid_df = dd.read_json("hf://datasets/allenai/c4/multilingual/c4-es-validation.*.json.gz")
1193
+ ```
1194
 
1195
+ ##### Using Git
1196
 
1197
  ```bash
1198
  git clone https://huggingface.co/datasets/allenai/c4
 
1212
  git lfs pull --include "multilingual/c4-nl.*.json.gz"
1213
  ```
1214
 
1215
+ ### Supported Tasks and Leaderboards
1216
 
1217
+ C4 and mC4 are mainly intended to pretrain language models and word representations.
1218
+
1219
+ ### Languages
1220
+
1221
+ The `en`, `en.noclean`, `en.noblocklist` and `realnewslike` variants are in English.
1222
+
1223
+ The other 108 languages are available and are reported in the table below.
1224
+
1225
+ Note that the languages that end with "-Latn" are simply romanized variants, i.e. written using the Latin script.
1226
+
1227
+
1228
+ | language code | language name |
1229
+ |:----------------|:---------------------|
1230
+ | af | Afrikaans |
1231
+ | am | Amharic |
1232
+ | ar | Arabic |
1233
+ | az | Azerbaijani |
1234
+ | be | Belarusian |
1235
+ | bg | Bulgarian |
1236
+ | bg-Latn | Bulgarian (Latin) |
1237
+ | bn | Bangla |
1238
+ | ca | Catalan |
1239
+ | ceb | Cebuano |
1240
+ | co | Corsican |
1241
+ | cs | Czech |
1242
+ | cy | Welsh |
1243
+ | da | Danish |
1244
+ | de | German |
1245
+ | el | Greek |
1246
+ | el-Latn | Greek (Latin) |
1247
+ | en | English |
1248
+ | eo | Esperanto |
1249
+ | es | Spanish |
1250
+ | et | Estonian |
1251
+ | eu | Basque |
1252
+ | fa | Persian |
1253
+ | fi | Finnish |
1254
+ | fil | Filipino |
1255
+ | fr | French |
1256
+ | fy | Western Frisian |
1257
+ | ga | Irish |
1258
+ | gd | Scottish Gaelic |
1259
+ | gl | Galician |
1260
+ | gu | Gujarati |
1261
+ | ha | Hausa |
1262
+ | haw | Hawaiian |
1263
+ | hi | Hindi |
1264
+ | hi-Latn | Hindi (Latin script) |
1265
+ | hmn | Hmong, Mong |
1266
+ | ht | Haitian |
1267
+ | hu | Hungarian |
1268
+ | hy | Armenian |
1269
+ | id | Indonesian |
1270
+ | ig | Igbo |
1271
+ | is | Icelandic |
1272
+ | it | Italian |
1273
+ | iw | former Hebrew |
1274
+ | ja | Japanese |
1275
+ | ja-Latn | Japanese (Latin) |
1276
+ | jv | Javanese |
1277
+ | ka | Georgian |
1278
+ | kk | Kazakh |
1279
+ | km | Khmer |
1280
+ | kn | Kannada |
1281
+ | ko | Korean |
1282
+ | ku | Kurdish |
1283
+ | ky | Kyrgyz |
1284
+ | la | Latin |
1285
+ | lb | Luxembourgish |
1286
+ | lo | Lao |
1287
+ | lt | Lithuanian |
1288
+ | lv | Latvian |
1289
+ | mg | Malagasy |
1290
+ | mi | Maori |
1291
+ | mk | Macedonian |
1292
+ | ml | Malayalam |
1293
+ | mn | Mongolian |
1294
+ | mr | Marathi |
1295
+ | ms | Malay |
1296
+ | mt | Maltese |
1297
+ | my | Burmese |
1298
+ | ne | Nepali |
1299
+ | nl | Dutch |
1300
+ | no | Norwegian |
1301
+ | ny | Nyanja |
1302
+ | pa | Punjabi |
1303
+ | pl | Polish |
1304
+ | ps | Pashto |
1305
+ | pt | Portuguese |
1306
+ | ro | Romanian |
1307
+ | ru | Russian |
1308
+ | ru-Latn | Russian (Latin) |
1309
+ | sd | Sindhi |
1310
+ | si | Sinhala |
1311
+ | sk | Slovak |
1312
+ | sl | Slovenian |
1313
+ | sm | Samoan |
1314
+ | sn | Shona |
1315
+ | so | Somali |
1316
+ | sq | Albanian |
1317
+ | sr | Serbian |
1318
+ | st | Southern Sotho |
1319
+ | su | Sundanese |
1320
+ | sv | Swedish |
1321
+ | sw | Swahili |
1322
+ | ta | Tamil |
1323
+ | te | Telugu |
1324
+ | tg | Tajik |
1325
+ | th | Thai |
1326
+ | tr | Turkish |
1327
+ | uk | Ukrainian |
1328
+ | und | Unknown language |
1329
+ | ur | Urdu |
1330
+ | uz | Uzbek |
1331
+ | vi | Vietnamese |
1332
+ | xh | Xhosa |
1333
+ | yi | Yiddish |
1334
+ | yo | Yoruba |
1335
+ | zh | Chinese |
1336
+ | zh-Latn | Chinese (Latin) |
1337
+ | zu | Zulu |
1338
+
1339
+ ## Dataset Structure
1340
+
1341
+ ### Data Instances
1342
+
1343
+ An example form the `en` config is:
1344
+
1345
+ ```
1346
+ {
1347
+ 'url': 'https://klyq.com/beginners-bbq-class-taking-place-in-missoula/',
1348
+ 'text': 'Beginners BBQ Class Taking Place in Missoula!\nDo you want to get better at making delicious BBQ? You will have the opportunity, put this on your calendar now. Thursday, September 22nd join World Class BBQ Champion, Tony Balay from Lonestar Smoke Rangers. He will be teaching a beginner level class for everyone who wants to get better with their culinary skills.\nHe will teach you everything you need to know to compete in a KCBS BBQ competition, including techniques, recipes, timelines, meat selection and trimming, plus smoker and fire information.\nThe cost to be in the class is $35 per person, and for spectators it is free. Included in the cost will be either a t-shirt or apron and you will be tasting samples of each meat that is prepared.',
1349
+ 'timestamp': '2019-04-25T12:57:54Z'
1350
+ }
1351
+ ```
1352
+
1353
+ ### Data Fields
1354
 
1355
+ The data have several fields:
1356
+
1357
+ - `url`: url of the source as a string
1358
+ - `text`: text content as a string
1359
+ - `timestamp`: timestamp as a string
1360
+
1361
+ ### Data Splits
1362
+
1363
+ Sizes for the variants in english:
1364
+
1365
+ | name | train |validation|
1366
+ |----------------|--------:|---------:|
1367
+ | en |364868892| 364608|
1368
+ | en.noblocklist |393391519| 393226|
1369
+ | en.noclean | ?| ?|
1370
+ | realnewslike | 13799838| 13863|
1371
+
1372
+ A train and validation split are also provided for the other languages, but lengths are still to be added.
1373
+
1374
+ ### Source Data
1375
+
1376
+ #### Initial Data Collection and Normalization
1377
+
1378
+ The C4 and mC4 datasets are collections text sourced from the public Common Crawl web scrape. It includes heuristics to extract only natural language (as opposed to boilerplate and other gibberish) in addition to extensive deduplication. You can find the code that has been used to build this dataset in [c4.py](https://github.com/tensorflow/datasets/blob/5952d3d60d60e1727786fa7a9a23d24bb463d4d6/tensorflow_datasets/text/c4.py) by Tensorflow Datasets.
1379
+
1380
+ C4 dataset was explicitly designed to be English only: any page that was not given a probability of at least 99% of being English by [langdetect](https://github.com/Mimino666/langdetect) was discarded.
1381
+
1382
+ To build mC4, the authors used [CLD3](https://github.com/google/cld3) to identify over 100 languages.
1383
+
1384
+ ### Licensing Information
1385
 
1386
  We are releasing this dataset under the terms of [ODC-BY](https://opendatacommons.org/licenses/by/1-0/). By using this, you are also bound by the [Common Crawl terms of use](https://commoncrawl.org/terms-of-use/) in respect of the content contained in the dataset.
1387
+
1388
+ ### Acknowledgements
1389
+
1390
+ Big ups to the good folks at [Common Crawl](https://commoncrawl.org) whose data made this possible ([consider donating](http://commoncrawl.org/donate/)!), to Google for creating the code that curates and filters the data, and to Huggingface, who had no issue with hosting these 3TB of data for public download!