Jeronymous commited on
Commit
ca27689
·
1 Parent(s): 3a22bc2

Change structure to welcome multilingual later

Browse files
README.md CHANGED
@@ -13,57 +13,17 @@ configs:
13
  - config_name: default
14
  data_files:
15
  - split: train
16
- path: 20231220/*
17
- - config_name: "20231220"
18
  data_files:
19
  - split: train
20
- path: 20231220/*
21
- - config_name: "20231220.1"
22
  data_files:
23
  - split: train
24
  path: 20231220/train-000000-of-000007.parquet
25
- - config_name: "20231220.2"
26
- data_files:
27
- - split: train
28
- path: 20231220/train-000001-of-000007.parquet
29
- - config_name: "20231220.3"
30
- data_files:
31
- - split: train
32
- path: 20231220/train-000002-of-000007.parquet
33
- - config_name: "20231220.4"
34
- data_files:
35
- - split: train
36
- path: 20231220/train-000003-of-000007.parquet
37
- - config_name: "20231220.5"
38
- data_files:
39
- - split: train
40
- path: 20231220/train-000004-of-000007.parquet
41
- - config_name: "20231220.6"
42
- data_files:
43
- - split: train
44
- path: 20231220/train-000005-of-000007.parquet
45
- - config_name: "20231220.7"
46
- data_files:
47
- - split: train
48
- path: 20231220/train-000006-of-000007.parquet
49
  dataset_info:
50
- - config_name: default
51
- features:
52
- - name: id
53
- dtype: int32
54
- - name: url
55
- dtype: string
56
- - name: title
57
- dtype: string
58
- - name: text
59
- dtype: string
60
- splits:
61
- - name: train
62
- num_bytes: 442573568
63
- num_examples: 650087
64
- download_size: 192941244
65
- dataset_size: 442573568
66
- - config_name: "20231220"
67
  features:
68
  - name: id
69
  dtype: int32
@@ -79,7 +39,7 @@ dataset_info:
79
  num_examples: 650087
80
  download_size: 192941244
81
  dataset_size: 442573568
82
- - config_name: "20231220.1"
83
  features:
84
  - name: id
85
  dtype: int32
@@ -95,126 +55,34 @@ dataset_info:
95
  num_examples: 91996
96
  download_size: 32850743
97
  dataset_size: 68635508
98
- - config_name: "20231220.2"
99
- features:
100
- - name: id
101
- dtype: int32
102
- - name: url
103
- dtype: string
104
- - name: title
105
- dtype: string
106
- - name: text
107
- dtype: string
108
- splits:
109
- - name: train
110
- num_bytes: 70148053
111
- num_examples: 118924
112
- download_size: 29931642
113
- dataset_size: 70148053
114
- - config_name: "20231220.3"
115
- features:
116
- - name: id
117
- dtype: int32
118
- - name: url
119
- dtype: string
120
- - name: title
121
- dtype: string
122
- - name: text
123
- dtype: string
124
- splits:
125
- - name: train
126
- num_bytes: 73099864
127
- num_examples: 119995
128
- download_size: 30711503
129
- dataset_size: 73099864
130
- - config_name: "20231220.4"
131
- features:
132
- - name: id
133
- dtype: int32
134
- - name: url
135
- dtype: string
136
- - name: title
137
- dtype: string
138
- - name: text
139
- dtype: string
140
- splits:
141
- - name: train
142
- num_bytes: 80580684
143
- num_examples: 110066
144
- download_size: 36351584
145
- dataset_size: 80580684
146
- - config_name: "20231220.5"
147
- features:
148
- - name: id
149
- dtype: int32
150
- - name: url
151
- dtype: string
152
- - name: title
153
- dtype: string
154
- - name: text
155
- dtype: string
156
- splits:
157
- - name: train
158
- num_bytes: 74534373
159
- num_examples: 102588
160
- download_size: 31438062
161
- dataset_size: 74534373
162
- - config_name: "20231220.6"
163
- features:
164
- - name: id
165
- dtype: int32
166
- - name: url
167
- dtype: string
168
- - name: title
169
- dtype: string
170
- - name: text
171
- dtype: string
172
- splits:
173
- - name: train
174
- num_bytes: 69491254
175
- num_examples: 98966
176
- download_size: 28958890
177
- dataset_size: 69491254
178
- - config_name: "20231220.7"
179
- features:
180
- - name: id
181
- dtype: int32
182
- - name: url
183
- dtype: string
184
- - name: title
185
- dtype: string
186
- - name: text
187
- dtype: string
188
- splits:
189
- - name: train
190
- num_bytes: 6083832
191
- num_examples: 7552
192
- download_size: 2698820
193
- dataset_size: 6083832
194
  ---
195
 
196
  # Plain text of French Wiktionary
197
-
198
- This dataset was created by [OpenLLM France](https://openllmfrance.org/) from the [Wikimedia dumps](https://dumps.wikimedia.org/other/enterprise_html/runs).
199
 
200
- It is a plain text version of pages from [fr.wiktionary.org/wiki](https://fr.wiktionary.org/wiki)
201
- without HTML tags nor wiki templates.
202
- The text just includes markdown syntax for headers, lists and tables.
203
- See [Notes on data formatting](#notes-on-data-formatting) for more details.
204
-
205
- * [Statistics](#statistics)
206
- * [Example use (python)](#example-use-python)
207
- * [Data fields](#data-fields)
208
- * [Notes on data formatting](#notes-on-data-formatting)
209
  * [License](#license)
210
  * [Aknowledgements](#aknowledgements)
211
  * [Citation](#citation)
212
 
 
 
 
 
 
 
 
 
 
 
213
 
214
- ## Statistics
215
 
216
  The amount of data for the latest dump (20231220) is:
217
- | **Unit** | **Size** |
218
  | :---------------|--------: |
219
  | # documents | 650 087 |
220
  | # paragraphs | 3 410 374 |
@@ -224,33 +92,28 @@ The amount of data for the latest dump (20231220) is:
224
  | size in memory | 422.1 MB |
225
  | size on disk | 184.0 MB |
226
 
227
- ## Example use (python)
228
 
229
  Load the full dataset:
230
 
231
  ```python
232
  import datasets
233
 
234
- ds = datasets.load_dataset("OpenLLM-France/wiktionary.fr")
235
  ```
236
 
237
  Load only a small subset:
238
  ```python
239
- import datasets
240
-
241
- i = 1 # Can be a number between 1 and 7
242
- subset = f"20231220.{i}"
243
-
244
- ds = datasets.load_dataset("OpenLLM-France/wiktionary.fr", subset)
245
  ```
246
 
247
  ## Data fields
248
 
249
  The data fields are the same among all configurations:
250
- - `id` (int): ID of the article.
251
- - `url` (`str`): URL of the article.
252
- - `title` (`str`): Title of the article.
253
- - `text` (`str`): Text content of the article.
254
 
255
  For example:
256
  ```
@@ -263,7 +126,7 @@ For example:
263
 
264
  ## Notes on data formatting
265
 
266
- see [OpenLLM-France/wikipedia.fr](https://huggingface.co/datasets/OpenLLM-France/wikipedia.fr)
267
 
268
  ## License
269
 
@@ -271,7 +134,9 @@ This dataset is distributed under the [Creative Commons Attribution-ShareAlike 4
271
 
272
  ## Aknowledgements
273
 
274
- This dataset was created by [Jérôme Louradour](https://huggingface.co/Jeronymous) on behalf of [OpenLLM France](https://openllmfrance.org/).
 
 
275
 
276
  Many thanks to the [Wikimedia Foundation](https://wikimediafoundation.org/)
277
  for providing the data and useful advices,
@@ -280,10 +145,10 @@ in particular Isaac Johnson, Albert Villanova and Rémy Gerbet.
280
  ## Citation
281
 
282
  ```
283
- @online{wikipedia_fr_dump,
284
- author = "OpenLLM France",
285
- title = "Plain text of French Wiktionary",
286
- url = "https://huggingface.co/datasets/OpenLLM-France/wiktionary.fr"
287
  }
288
  ```
289
 
 
13
  - config_name: default
14
  data_files:
15
  - split: train
16
+ path: "*/20231220/*.parquet"
17
+ - config_name: "fr"
18
  data_files:
19
  - split: train
20
+ path: "fr/20231220/*.parquet"
21
+ - config_name: "sample"
22
  data_files:
23
  - split: train
24
  path: 20231220/train-000000-of-000007.parquet
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
25
  dataset_info:
26
+ - config_name: "fr"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
27
  features:
28
  - name: id
29
  dtype: int32
 
39
  num_examples: 650087
40
  download_size: 192941244
41
  dataset_size: 442573568
42
+ - config_name: "sample"
43
  features:
44
  - name: id
45
  dtype: int32
 
55
  num_examples: 91996
56
  download_size: 32850743
57
  dataset_size: 68635508
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
58
  ---
59
 
60
  # Plain text of French Wiktionary
 
 
61
 
62
+ * [Dataset Description](#dataset-description)
63
+ * [Size](#size)
64
+ * [Example use (python)](#example-use-python)
65
+ * [Data fields](#data-fields)
66
+ * [Notes on data formatting](#notes-on-data-formatting)
 
 
 
 
67
  * [License](#license)
68
  * [Aknowledgements](#aknowledgements)
69
  * [Citation](#citation)
70
 
71
+ ## Dataset Description
72
+
73
+ This dataset is a plain text version of pages from [wiktionary.org](https://wiktionary.org/) in French language.
74
+ The text is without HTML tags nor wiki templates.
75
+ It just includes markdown syntax for headers, lists and tables.
76
+ See [Notes on data formatting](#notes-on-data-formatting) for more details.
77
+
78
+ It was created by [LINAGORA](https://labs.linagora.com/) and [OpenLLM France](https://openllmfrance.org/)
79
+ from the [Wikimedia dumps](https://dumps.wikimedia.org/other/enterprise_html/runs),
80
+ using code in [https://github.com/OpenLLM-France/wikiplaintext](https://github.com/OpenLLM-France/wikiplaintext).
81
 
82
+ ### Size
83
 
84
  The amount of data for the latest dump (20231220) is:
85
+ | | **French** |
86
  | :---------------|--------: |
87
  | # documents | 650 087 |
88
  | # paragraphs | 3 410 374 |
 
92
  | size in memory | 422.1 MB |
93
  | size on disk | 184.0 MB |
94
 
95
+ ### Example use (python)
96
 
97
  Load the full dataset:
98
 
99
  ```python
100
  import datasets
101
 
102
+ ds = datasets.load_dataset("OpenLLM-France/wiktionary.fr", streaming=True, split="train")
103
  ```
104
 
105
  Load only a small subset:
106
  ```python
107
+ ds = datasets.load_dataset("OpenLLM-France/wiktionary.fr", "sample", split="train")
 
 
 
 
 
108
  ```
109
 
110
  ## Data fields
111
 
112
  The data fields are the same among all configurations:
113
+ - `id` (int): ID of the page.
114
+ - `url` (`str`): URL of the page.
115
+ - `title` (`str`): Title of the page.
116
+ - `text` (`str`): Text content of the page.
117
 
118
  For example:
119
  ```
 
126
 
127
  ## Notes on data formatting
128
 
129
+ see [OpenLLM-France/wikipedia.fr](https://huggingface.co/datasets/OpenLLM-France/wikipedia)
130
 
131
  ## License
132
 
 
134
 
135
  ## Aknowledgements
136
 
137
+ This dataset was created by [Jérôme Louradour](https://huggingface.co/Jeronymous) on behalf of
138
+ [LINAGORA](https://labs.linagora.com/) and
139
+ [OpenLLM France](https://openllmfrance.org/).
140
 
141
  Many thanks to the [Wikimedia Foundation](https://wikimediafoundation.org/)
142
  for providing the data and useful advices,
 
145
  ## Citation
146
 
147
  ```
148
+ @online{wiktionary_fr_dump,
149
+ author = "Jérôme Louradour, OpenLLM-France, LINAGORA Labs",
150
+ title = "Plain text of Wiktionary",
151
+ url = "https://huggingface.co/datasets/OpenLLM-France/wiktionary"
152
  }
153
  ```
154
 
{20231220 → fr/20231220}/train-000000-of-000007.parquet RENAMED
File without changes
{20231220 → fr/20231220}/train-000001-of-000007.parquet RENAMED
File without changes
{20231220 → fr/20231220}/train-000002-of-000007.parquet RENAMED
File without changes
{20231220 → fr/20231220}/train-000003-of-000007.parquet RENAMED
File without changes
{20231220 → fr/20231220}/train-000004-of-000007.parquet RENAMED
File without changes
{20231220 → fr/20231220}/train-000005-of-000007.parquet RENAMED
File without changes
{20231220 → fr/20231220}/train-000006-of-000007.parquet RENAMED
File without changes