File size: 11,585 Bytes
8c32667
 
 
 
 
 
dd09f94
8c32667
dd09f94
8c32667
 
 
 
 
 
 
 
 
 
ec1407a
 
 
1cc383c
3a412ae
 
42289fe
 
 
 
 
 
 
 
 
1cc383c
 
42289fe
 
 
 
 
 
d1743c1
 
 
42289fe
 
 
 
 
 
 
 
 
 
 
 
 
d1743c1
 
 
42289fe
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1cc383c
 
 
42289fe
 
 
d1743c1
 
 
42289fe
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d1743c1
 
 
42289fe
 
28f21e0
 
 
 
 
8c32667
 
 
 
 
 
 
76b3800
8c32667
 
76b3800
 
 
8c32667
 
 
 
 
 
 
 
 
 
 
 
 
089fab5
8c32667
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
089fab5
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
---
annotations_creators:
- crowdsourced
- machine-generated
language_creators:
- crowdsourced
language:
- fr
license:
- unknown
multilinguality:
- monolingual
size_categories:
- 10K<n<100K
source_datasets:
- original
task_categories:
- text-classification
task_ids:
- intent-classification
- semantic-similarity-classification
- sentiment-classification
pretty_name: FLUE
tags:
- Word Sense Disambiguation for Verbs
dataset_info:
- config_name: CLS
  features:
  - name: text
    dtype: string
  - name: label
    dtype:
      class_label:
        names:
          '0': negative
          '1': positive
  - name: idx
    dtype: int32
  splits:
  - name: train
    num_bytes: 3853279
    num_examples: 5997
  - name: test
    num_bytes: 3852344
    num_examples: 5999
  download_size: 314687066
  dataset_size: 7705623
- config_name: PAWS-X
  features:
  - name: sentence1
    dtype: string
  - name: sentence2
    dtype: string
  - name: label
    dtype: int32
  - name: idx
    dtype: int32
  splits:
  - name: validation
    num_bytes: 522013
    num_examples: 1988
  - name: test
    num_bytes: 526953
    num_examples: 2000
  - name: train
    num_bytes: 13096677
    num_examples: 49399
  download_size: 30282057
  dataset_size: 14145643
- config_name: XNLI
  features:
  - name: premise
    dtype: string
  - name: hypo
    dtype: string
  - name: label
    dtype:
      class_label:
        names:
          '0': contradiction
          '1': entailment
          '2': neutral
  - name: idx
    dtype: int32
  splits:
  - name: validation
    num_bytes: 520022
    num_examples: 2490
  - name: test
    num_bytes: 1048999
    num_examples: 5010
  - name: train
    num_bytes: 87373154
    num_examples: 392702
  download_size: 483963712
  dataset_size: 88942175
- config_name: WSD-V
  features:
  - name: sentence
    sequence: string
  - name: pos_tags
    sequence: string
  - name: lemmas
    sequence: string
  - name: fine_pos_tags
    sequence: string
  - name: disambiguate_tokens_ids
    sequence: int32
  - name: disambiguate_labels
    sequence: string
  - name: idx
    dtype: string
  splits:
  - name: train
    num_bytes: 206869215
    num_examples: 269821
  - name: test
    num_bytes: 2722232
    num_examples: 3121
  download_size: 38303600
  dataset_size: 209591447
config_names:
- CLS
- PAWS-X
- WSD-V
- XNLI
---

# Dataset Card for FLUE

## Table of Contents
- [Dataset Description](#dataset-description)
  - [Dataset Summary](#dataset-summary)
  - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
  - [Languages](#languages)
- [Dataset Structure](#dataset-structure)
  - [Data Instances](#data-instances)
  - [Data Fields](#data-fields)
  - [Data Splits](#data-splits)
- [Dataset Creation](#dataset-creation)
  - [Curation Rationale](#curation-rationale)
  - [Source Data](#source-data)
  - [Annotations](#annotations)
  - [Personal and Sensitive Information](#personal-and-sensitive-information)
- [Considerations for Using the Data](#considerations-for-using-the-data)
  - [Social Impact of Dataset](#social-impact-of-dataset)
  - [Discussion of Biases](#discussion-of-biases)
  - [Other Known Limitations](#other-known-limitations)
- [Additional Information](#additional-information)
  - [Dataset Curators](#dataset-curators)
  - [Licensing Information](#licensing-information)
  - [Citation Information](#citation-information)
  - [Contributions](#contributions)

## Dataset Description

- **Homepage:** [homepage](https://github.com/getalp/Flaubert/tree/master/flue)
- **Repository:**[github](https://github.com/getalp/Flaubert/tree/master/flue)
- **Paper:**[paper](https://arxiv.org/abs/1912.05372)
- **Leaderboard:**[leaderboard](https://github.com/getalp/Flaubert/tree/master/flue/leaderboard)
- **Point of Contact:**[Hang Le](thi-phuong-hang.le@univ-grenoble-alpes.fr)

### Dataset Summary

FLUE is an evaluation setup for French NLP systems similar to the popular GLUE benchmark. The goal is to enable further reproducible experiments in the future and to share models and progress on the French language. The tasks and data are obtained from existing works, please refer to our Flaubert paper for a complete list of references.

### Supported Tasks and Leaderboards

The supported tasks are: Text Classification, Paraphrasing, Natural Language Inference, Constituency Parsing, Dependency Parsing, Verb Sense Disambiguation and Noun Sense Disambiguation

### Languages

The datasets are all in French.

## Dataset Structure

### Text Classification (CLS)

This is a binary classification task. It consists in classifying Amazon reviews for three product categories: books, DVD, and music. Each sample contains a review text and the associated rating from 1 to 5 stars. Reviews rated above 3 is labeled as positive, and those rated less than 3 is labeled as negative.

#### Data Instances

An instance looks like:

```
{
    'idx': 1,
    'label': 0,
    'text': 'Bilan plus que mitigé pour cet album fourre-tout qui mêle quelques bonnes idées (les parodies d\'oeuvres d\'art) et des scènetes qui ne font que faire écho paresseusement aux précédents albums. Uderzo n\'a pas pris de risque pour cet album, mais, au vu des précédents, on se dit que c\'est peut-être un moindre mal ... L\'album semble n\'avoir été fait que pour permettre à Uderzo de rappeler avec une insistance suspecte qu\'il est bien l\'un des créateurs d\'Astérix (comme lorsqu\'il se met en scène lui même dans la BD) et de traiter ses critiques d\' "imbéciles" dans une préface un rien aigrie signée "Astérix". Préface dans laquelle Uderzo feint de croire que ce qu\'on lui reproche est d\'avoir fait survivre Asterix à la disparition de Goscinny (reproche naturellement démenti par la fidélité des lecteurs - démonstration imparable !). On aurait tant aimé qu\'Uderzo accepte de s\'entourer d\'un scénariste compétent et respectueux de l\'esprit Goscinnien (cela doit se trouver !) et nous propose des albums plus ambitieux ...'
}
```

#### Data Fields

The dataset is composed of two fields:
- **text**: the field that represents the text to classify.
- **label**: the sentiment represented by the text, here **positive** or **negative**.

#### Data Splits

The train and test sets are balanced, including around 1k positive and 1k negative reviews for a total of 2k reviews in each dataset. We take the French portion to create the binary text classification task in FLUE and report the accuracy on the test set.

### Paraphrasing (PAWS-X)

The task consists in identifying whether the two sentences in a pair are semantically equivalent or not.

#### Data Instances

An instance looks like:

```
{
    'idx': 1,
    'label': 0,
    'sentence1': "À Paris, en octobre 1560, il rencontra secrètement l'ambassadeur d'Angleterre, Nicolas Throckmorton, lui demandant un passeport pour retourner en Angleterre en passant par l'Écosse.",
    'sentence2': "En octobre 1560, il rencontra secrètement l'ambassadeur d'Angleterre, Nicolas Throckmorton, à Paris, et lui demanda un passeport pour retourner en Écosse par l'Angleterre."
}
```

#### Data Fields

The dataset is compososed of three fields:
- **sentence1**: The first sentence of an example
- **sentence2**: The second sentence of an example
- **lalel**: **0** if the two sentences are not paraphrasing each other, **1** otherwise.

#### Data Splits

The train set includes 49.4k examples, the dev and test sets each comprises nearly 2k examples. We take the related datasets for French to perform the paraphrasing task and report the accuracy on the test set.

### Natural Language Inference (XNLI)

The Natural Language Inference (NLI) task, also known as recognizing textual entailment (RTE), is to determine whether a premise entails, contradicts or neither entails nor contradicts a hypothesis. We take the French part of the XNLI corpus to form the development and test sets for the NLI task in FLUE.

#### Data Instances

An instance looks like:

```
{
    'idx': 1,
    'label': 2,
    'hypo': 'Le produit et la géographie sont ce qui fait travailler la crème de la crème .',
    'premise': "L' écrémage conceptuel de la crème a deux dimensions fondamentales : le produit et la géographie ."
}
```

#### Data Fields

The dataset is composed of three fields:
- **premise**: Premise sentence.
- **hypo**: Hypothesis sentence.
- **label**: **contradiction** if the two sentences are contradictory, **entailment** if the two sentences entails, **neutral** if they neither entails or contradict each other.

#### Data Splits

The train set includes 392.7k examples, the dev and test sets comprises 2.5k and 5k examples respectively. We take the related datasets for French to perform the NLI task and report the accuracy on the test set.

### Word Sense Disambiguation for Verbs (WSD-V)

The FrenchSemEval (FSE) dataset aims to evaluate the Word Sense Disambiguation for Verbs task for the French language. Extracted from Wiktionary.

#### Data Instances

An instance looks like:

```
{
    'idx': 'd000.s001',
    'sentence': ['"', 'Ce', 'ne', 'fut', 'pas', 'une', 'révolution', '2.0', ',', 'ce', 'fut', 'une', 'révolution', 'de', 'rue', '.'],
    'fine_pos_tags': [27, 26, 18, 13, 18, 0, 6, 22, 27, 26, 13, 0, 6, 4, 6, 27],
    'lemmas': ['"', 'ce', 'ne', 'être', 'pas', 'un', 'révolution', '2.0', ',', 'ce', 'être', 'un', 'révolution', 'de', 'rue', '.'],
    'pos_tags': [13, 11, 14, 0, 14, 9, 15, 4, 13, 11, 0, 9, 15, 7, 15, 13],
    'disambiguate_labels': ['__ws_1_2.0__adj__1'],
    'disambiguate_tokens_ids': [7],
}
```

#### Data Fields

The dataset is composed of six fields:
- **sentence**: The sentence to process split in tokens.
- **pos_tags**: The corresponding POS tags for each tokens.
- **lemmas**: The corresponding lemma for each tokens.
- **fine_pos_tags**: Fined (more specific) POS tags for each tokens.
- **disambiguate_tokens_ids**: The ID of the token in the sentence to disambiguate.
- **disambiguate_labels**: The label in the form of **sentenceID __ws_sentence-number_token__pos__number-of-time-the-token-appeared-across-all-the-sentences** (i.e. **d000.s404.t000 __ws_2_agir__verb__1**). 

#### Data Splits

The train set includes 269821 examples, the test set includes 3121 examples.

## Considerations for Using the Data

### Social Impact of Dataset

The goal is to enable further reproducible experiments in the future and to share models and progress on the French language.

## Additional Information

### Licensing Information

The licenses are:
- The licensing status of the data, especially the news source text, is unknown for CLS
- *The dataset may be freely used for any purpose, although acknowledgement of Google LLC ("Google") as the data source would be appreciated. The dataset is provided "AS IS" without any warranty, express or implied. Google disclaims all liability for any damages, direct or indirect, resulting from the use of the dataset.* for PAWS-X
- CC BY-NC 4.0 for XNLI
- The licensing status of the data, especially the news source text, is unknown for Verb Sense Disambiguation 

### Citation Information

```
@misc{le2019flaubert,
    title={FlauBERT: Unsupervised Language Model Pre-training for French},
    author={Hang Le and Loïc Vial and Jibril Frej and Vincent Segonne and Maximin Coavoux and Benjamin Lecouteux and Alexandre Allauzen and Benoît Crabbé and Laurent Besacier and Didier Schwab},
    year={2019},
    eprint={1912.05372},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}
```
### Contributions

Thanks to [@jplu](https://github.com/jplu) for adding this dataset.