Datasets:
Jeronymous
commited on
Commit
·
78d6201
1
Parent(s):
2bfc158
Update README
Browse files
README.md
CHANGED
@@ -1,3 +1,281 @@
|
|
1 |
-
---
|
2 |
-
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language:
|
3 |
+
- en
|
4 |
+
- fr
|
5 |
+
- de
|
6 |
+
- es
|
7 |
+
- it
|
8 |
+
license:
|
9 |
+
- cc-by-sa-4.0
|
10 |
+
task_categories:
|
11 |
+
- text-generation
|
12 |
+
- fill-mask
|
13 |
+
task_ids:
|
14 |
+
- language-modeling
|
15 |
+
- masked-language-modeling
|
16 |
+
configs:
|
17 |
+
- config_name: default
|
18 |
+
data_files:
|
19 |
+
- split: train
|
20 |
+
path: "*/20240201/*parquet"
|
21 |
+
- config_name: "en"
|
22 |
+
data_files:
|
23 |
+
- split: train
|
24 |
+
path: "en/20240201/*parquet"
|
25 |
+
- config_name: "fr"
|
26 |
+
data_files:
|
27 |
+
- split: train
|
28 |
+
path: "fr/20240201/*parquet"
|
29 |
+
- config_name: "de"
|
30 |
+
data_files:
|
31 |
+
- split: train
|
32 |
+
path: "de/20240201/*parquet"
|
33 |
+
- config_name: "es"
|
34 |
+
data_files:
|
35 |
+
- split: train
|
36 |
+
path: "es/20240201/*parquet"
|
37 |
+
- config_name: "it"
|
38 |
+
data_files:
|
39 |
+
- split: train
|
40 |
+
path: "it/20240201/*parquet"
|
41 |
+
dataset_info:
|
42 |
+
- config_name: "en"
|
43 |
+
features:
|
44 |
+
- name: id
|
45 |
+
dtype: int32
|
46 |
+
- name: url
|
47 |
+
dtype: string
|
48 |
+
- name: title
|
49 |
+
dtype: string
|
50 |
+
- name: text
|
51 |
+
dtype: string
|
52 |
+
splits:
|
53 |
+
- name: train
|
54 |
+
num_bytes: 28019684686
|
55 |
+
num_examples: 6893170
|
56 |
+
download_size: 15544920302
|
57 |
+
dataset_size: 28019684686
|
58 |
+
- config_name: "fr"
|
59 |
+
features:
|
60 |
+
- name: id
|
61 |
+
dtype: int32
|
62 |
+
- name: url
|
63 |
+
dtype: string
|
64 |
+
- name: title
|
65 |
+
dtype: string
|
66 |
+
- name: text
|
67 |
+
dtype: string
|
68 |
+
splits:
|
69 |
+
- name: train
|
70 |
+
num_bytes: 10734253565
|
71 |
+
num_examples: 2647717
|
72 |
+
download_size: 5990349749
|
73 |
+
dataset_size: 10734253565
|
74 |
+
- config_name: "de"
|
75 |
+
features:
|
76 |
+
- name: id
|
77 |
+
dtype: int32
|
78 |
+
- name: url
|
79 |
+
dtype: string
|
80 |
+
- name: title
|
81 |
+
dtype: string
|
82 |
+
- name: text
|
83 |
+
dtype: string
|
84 |
+
splits:
|
85 |
+
- name: train
|
86 |
+
num_bytes: 11947016839
|
87 |
+
num_examples: 2877387
|
88 |
+
download_size: 6811415105
|
89 |
+
dataset_size: 11947016839
|
90 |
+
- config_name: "es"
|
91 |
+
features:
|
92 |
+
- name: id
|
93 |
+
dtype: int32
|
94 |
+
- name: url
|
95 |
+
dtype: string
|
96 |
+
- name: title
|
97 |
+
dtype: string
|
98 |
+
- name: text
|
99 |
+
dtype: string
|
100 |
+
splits:
|
101 |
+
- name: train
|
102 |
+
num_bytes: 7636098849
|
103 |
+
num_examples: 1947242
|
104 |
+
download_size: 4346936003
|
105 |
+
dataset_size: 7636098849
|
106 |
+
- config_name: "it"
|
107 |
+
features:
|
108 |
+
- name: id
|
109 |
+
dtype: int32
|
110 |
+
- name: url
|
111 |
+
dtype: string
|
112 |
+
- name: title
|
113 |
+
dtype: string
|
114 |
+
- name: text
|
115 |
+
dtype: string
|
116 |
+
splits:
|
117 |
+
- name: train
|
118 |
+
num_bytes: 6544616305
|
119 |
+
num_examples: 1870387
|
120 |
+
download_size: 3742980452
|
121 |
+
dataset_size: 6544616305
|
122 |
+
---
|
123 |
+
|
124 |
+
# Plain text of Wikipedia
|
125 |
+
|
126 |
+
* [Dataset Description](#dataset-description)
|
127 |
+
* [Size](#size)
|
128 |
+
* [Example use (python)](#example-use-python)
|
129 |
+
* [Data fields](#data-fields)
|
130 |
+
* [Notes on data formatting](#notes-on-data-formatting)
|
131 |
+
* [License](#license)
|
132 |
+
* [Aknowledgements](#aknowledgements)
|
133 |
+
* [Citation](#citation)
|
134 |
+
|
135 |
+
## Dataset Description
|
136 |
+
|
137 |
+
This dataset is a plain text version of pages from `wikipedia.org` spaces for several languages
|
138 |
+
([English](https://en.wikipedia.org/wiki),
|
139 |
+
[German](https://de.wikipedia.org/wiki),
|
140 |
+
[French](https://fr.wikipedia.org/wiki),
|
141 |
+
[Spanish](https://es.wikipedia.org/wiki),
|
142 |
+
[Italian](https://it.wikipedia.org/wiki)).
|
143 |
+
The text is without HTML tags nor wiki templates.
|
144 |
+
It just includes markdown syntax for headers, lists and tables.
|
145 |
+
See [Notes on data formatting](#notes-on-data-formatting) for more details.
|
146 |
+
|
147 |
+
It was created by [LINAGORA](https://labs.linagora.com/) and [OpenLLM France](https://openllmfrance.org/)
|
148 |
+
from the [Wikimedia dumps](https://dumps.wikimedia.org/other/enterprise_html/runs),
|
149 |
+
using code in [https://github.com/OpenLLM-France/wikiplaintext](https://github.com/OpenLLM-France/wikiplaintext).
|
150 |
+
|
151 |
+
### Size
|
152 |
+
|
153 |
+
The amount of data for the latest dump (20240201) are for the several languages:
|
154 |
+
| | **English (en)** | **German (de)** | **French (fr)** | **Spanish (es)** | **Italian (it)** |
|
155 |
+
| :---------------|--------: |--------: |--------: |--------: |--------: |
|
156 |
+
| **# documents** | 6 893 170 | 2 877 387 | 2 647 717 | 1 947 242 | 1 870 387 |
|
157 |
+
| **# paragraphs** | 59 010 400 | 23 686 448 | 24 943 988 | 16 354 822 | 15 774 432 |
|
158 |
+
| **# lines** | 241 540 701 | 90 393 406 | 97 295 694 | 56 476 769 | 61 007 593 |
|
159 |
+
| **# words** | 4 772 115 884 | 1 730 527 517 | 1 758 293 925 | 1 263 717 667 | 1 078 659 862 |
|
160 |
+
| **# characters** | 27 172 681 137 | 11 468 147 020 | 10 141 322 856 | 7 310 434 534 | 6 315 423 707 |
|
161 |
+
| **size in memory** | 26 721.7 MB | 11 393.6 MB | 10 237.0 MB | 7 282.4 MB | 6 241.4 MB |
|
162 |
+
| **size on disk** | 14 824.8 MB | 6 495.9 MB | 5 712.8 MB | 4 145.6 MB | 3 569.6 MB |
|
163 |
+
|
164 |
+
|
165 |
+
### Example use (python)
|
166 |
+
|
167 |
+
Load the full dataset:
|
168 |
+
|
169 |
+
```python
|
170 |
+
import datasets
|
171 |
+
|
172 |
+
ds = datasets.load_dataset("OpenLLM-France/wikipedia",
|
173 |
+
streaming=True,
|
174 |
+
split="train"
|
175 |
+
)
|
176 |
+
```
|
177 |
+
|
178 |
+
Load the dataset for a given language (French in this example):
|
179 |
+
```python
|
180 |
+
ds = datasets.load_dataset("OpenLLM-France/wikipedia", "fr",
|
181 |
+
streaming=True,
|
182 |
+
split="train"
|
183 |
+
)
|
184 |
+
```
|
185 |
+
|
186 |
+
### Data fields
|
187 |
+
|
188 |
+
The data fields are the same among all configurations:
|
189 |
+
- `id` (`int`): ID of the article.
|
190 |
+
- `url` (`str`): URL of the article.
|
191 |
+
- `title` (`str`): Title of the article.
|
192 |
+
- `text` (`str`): Text content of the article.
|
193 |
+
|
194 |
+
For example:
|
195 |
+
```
|
196 |
+
{'id': 847027,
|
197 |
+
'url': 'https://fr.wikipedia.org/wiki/D%C3%A9partement_de_Rinconada',
|
198 |
+
'title': 'Département de Rinconada',
|
199 |
+
'text': "# Département de Rinconada\n\nLe département de Rinconada est une des 16 subdivisions de la province de Jujuy, en Argentine. Son chef-lieu est la ville de Rinconada.\nLe département a une superficie de 6 407 km². Sa population extrêmement clairsemée s'élevait à 2 298 habitants en 2001, soit une densité de 0,3 hab./km².\n\n# Département de Rinconada\n## Localités\n\nOutre le chef-lieu d'arrondissement, Rinconada, il faut citer :\n* Mina Pirquitas, localité la plus élevée d'Argentine (4 271 mètres d'altitude).\nC'est sur le territoire de ce département qu'est situé le site Ramsar des Lagunas de Vilama."
|
200 |
+
}
|
201 |
+
```
|
202 |
+
|
203 |
+
### Notes on data formatting
|
204 |
+
|
205 |
+
#### Special characters
|
206 |
+
|
207 |
+
Superscripts and subscripts are kept as unicode characters when possible. e.g. `XIIIᵉ siècle`, `3 000 m²`, `P₂O₇⁴⁻`.
|
208 |
+
|
209 |
+
Unbreakable spaces are kept as unicode characters
|
210 |
+
(in python, use `text.replace("\u00A0", " ")` to replace them with normal spaces).
|
211 |
+
|
212 |
+
#### Markdown syntax
|
213 |
+
|
214 |
+
Tables are kept as markdown tables.
|
215 |
+
For instance:
|
216 |
+
<div style="border: 2px solid black; padding: 10px;"><code>
|
217 |
+
| Français | Espéranto | IPA |<br/>
|
218 |
+
| Salut, bonjour | Saluton | [sa.ˈlu.ton] |<br/>
|
219 |
+
| Oui | Jes | [ˈjes] |<br/>
|
220 |
+
| Non | Ne | [ˈne] |<br/>
|
221 |
+
| Bonsoir | Bonan vesperon | [ˈbo.nan ves.ˈpe.ron] |<br/>
|
222 |
+
| Bonne nuit | Bonan nokton | [ˈbo.nan ˈnok.ton] |
|
223 |
+
</code></div>
|
224 |
+
|
225 |
+
Lists are kept as markdown lists.
|
226 |
+
For instance:
|
227 |
+
<div style="border: 2px solid black; padding: 10px;"><code>
|
228 |
+
* 1 000 personnes ont l'espéranto comme langue maternelle ;<br/>
|
229 |
+
* 10 000 personnes parlent l'espéranto avec un niveau proche d'une langue maternelle ;<br/>
|
230 |
+
* 100 000 personnes parlent couramment l'espéranto ;
|
231 |
+
</code></div>
|
232 |
+
|
233 |
+
Headers are kept as markdown headers, and all the levels of headers are kept before each paragraph.
|
234 |
+
The first header always corresponds to the page title.
|
235 |
+
For instance:
|
236 |
+
<div style="border: 2px solid black; padding: 10px;"><code>
|
237 |
+
# Espéranto<br />
|
238 |
+
<br />
|
239 |
+
L'espéranto est une langue construite internationale utilisée comme langue véhiculaire par des personnes […]<br />
|
240 |
+
Fondée sur une grammaire régulière sans exception, l'espéranto est une langue globalement agglutinante où[…]<br />
|
241 |
+
C’est en 1887 que Louis-Lazare Zamenhof, sous le pseudonyme Doktoro Esperanto (Docteur « Espérant ») qui […]<br />
|
242 |
+
L’Association universelle d’espéranto, fondée en 1908, est en relation officielle avec l’Organisation des[…]<br />
|
243 |
+
<br />
|
244 |
+
# Espéranto<br />
|
245 |
+
## Définition<br />
|
246 |
+
### Nom<br />
|
247 |
+
<br />
|
248 |
+
Le pseudonyme « Doktoro Esperanto » (Docteur « Espérant »), utilisé par Zamenhof pour présenter son proje[…]<br />
|
249 |
+
<br />
|
250 |
+
# Espéranto<br />
|
251 |
+
## Définition<br />
|
252 |
+
### Nom<br />
|
253 |
+
#### Utilisation du mot espéranto en tant que métaphore<br />
|
254 |
+
<br />
|
255 |
+
Le nom espéranto fonctionne comme un nom propre quand il désigne la langue même, mais est parfois utilisé[…]<br />
|
256 |
+
Dans le domaine de l'informatique, Java fut qualifié d'« espéranto des langages de programmation », en pa[…]
|
257 |
+
</code></div>
|
258 |
+
|
259 |
+
## License
|
260 |
+
|
261 |
+
This dataset is distributed under the [Creative Commons Attribution-ShareAlike 4.0 International License](https://creativecommons.org/licenses/by-sa/4.0/).
|
262 |
+
|
263 |
+
## Aknowledgements
|
264 |
+
|
265 |
+
This dataset was created by [Jérôme Louradour](https://huggingface.co/Jeronymous) on behalf of
|
266 |
+
[LINAGORA](https://labs.linagora.com/) and
|
267 |
+
[OpenLLM France](https://openllmfrance.org/).
|
268 |
+
|
269 |
+
Many thanks to the [Wikimedia Foundation](https://wikimediafoundation.org/)
|
270 |
+
for providing the data and useful advices,
|
271 |
+
in particular Isaac Johnson, Albert Villanova and Rémy Gerbet.
|
272 |
+
|
273 |
+
## Citation
|
274 |
+
|
275 |
+
```
|
276 |
+
@online{wikipedia_dump,
|
277 |
+
author = "OpenLLM-France, LINAGORA Labs",
|
278 |
+
title = "Plain text Wikipedia",
|
279 |
+
url = "https://huggingface.co/datasets/OpenLLM-France/wikipedia"
|
280 |
+
}
|
281 |
+
```
|