bourdoiscatie commited on
Commit
ca7c5f7
1 Parent(s): f0a18ad

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +80 -0
README.md ADDED
@@ -0,0 +1,80 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - fr
4
+ license: cc-by-sa-4.0
5
+ size_categories:
6
+ - 10K<n<100K
7
+ task_categories:
8
+ - fill-mask
9
+ ---
10
+
11
+ # orange_sum_fr_prompt_fill_mask
12
+
13
+ ## Summary
14
+
15
+ **orange_sum_fr_prompt_fill_mask** is a subset of the [**Dataset of French Prompts (DFP)**]().
16
+ It contains **48,069** rows that can be used for a fill mask task.
17
+ The original data (without prompts) comes from the dataset [orange_sum](https://huggingface.co/datasets/orange_sum) by Eddine et al.
18
+ A list of prompts (see below) was then applied in order to build the input and target columns and thus obtain the same format as the [xP3](https://huggingface.co/datasets/bigscience/xP3) dataset by Muennighoff et al.
19
+
20
+
21
+ ## Prompts used
22
+ ### List
23
+ 24 prompts were created for this dataset. The logic applied consists in proposing prompts in the indicative tense, in the form of tutoiement and in the form of vouvoiement.
24
+
25
+ ```
26
+ 'Remplacer le <mask> dans le texte suivant par le mot le plus vraisemblable : '+text,
27
+ 'Remplace le <mask> dans le texte suivant par le mot le plus vraisemblable : '+text,
28
+ 'Remplacez le <mask> dans le texte suivant par le mot le plus vraisemblable : '+text,
29
+ 'Remplacer le <mask> dans le texte suivant par le mot le plus probable : '+text,
30
+ 'Remplace le <mask> dans le texte suivant par le mot le plus probable : '+text,
31
+ 'Remplacez le <mask> dans le texte suivant par le mot le plus probable : '+text,
32
+ 'Substituer le <mask> dans le texte suivant par le mot le plus vraisemblable : '+text,
33
+ 'Substitue le <mask> dans le texte suivant par le mot le plus vraisemblable : '+text,
34
+ 'Substituez le <mask> dans le texte suivant par le mot le plus vraisemblable : '+text,
35
+ 'Substituer le <mask> dans le texte suivant par le mot le plus probable : '+text,
36
+ 'Substitue le <mask> dans le texte suivant par le mot le plus probable : '+text,
37
+ 'Substituez le <mask> dans le texte suivant par le mot le plus probable : '+text,
38
+ 'Changer le <mask> dans le texte suivant par le mot le plus vraisemblable : '+text,
39
+ 'Change le <mask> dans le texte suivant par le mot le plus vraisemblable : '+text,
40
+ 'Changez le <mask> dans le texte suivant par le mot le plus vraisemblable : '+text,
41
+ 'Changer le <mask> dans le texte suivant par le mot le plus probable : '+text,
42
+ 'Change le <mask> dans le texte suivant par le mot le plus probable : '+text,
43
+ 'Changez le <mask> dans le texte suivant par le mot le plus probable : '+text,
44
+ 'Transformer le <mask> dans le texte suivant par le mot le plus vraisemblable : '+text,
45
+ 'Transforme le <mask> dans le texte suivant par le mot le plus vraisemblable : '+text,
46
+ 'Transformez le <mask> dans le texte suivant par le mot le plus vraisemblable : '+text,
47
+ 'Transformer le <mask> dans le texte suivant par le mot le plus probable : '+text,
48
+ 'Transforme le <mask> dans le texte suivant par le mot le plus probable : '+text,
49
+ 'Transformez le <mask> dans le texte suivant par le mot le plus probable : '+text,
50
+ ```
51
+
52
+
53
+ # Splits
54
+ - train with 48,069 samples
55
+ - no valid and test splits in the original data
56
+
57
+
58
+ # How to use?
59
+ ```
60
+ from datasets import load_dataset
61
+ dataset = load_dataset("CATIE-AQ/orange_sum_fr_prompt_fill_mask")
62
+ ```
63
+
64
+ # Citation
65
+ ## Original data
66
+ > @article{eddine2020barthez,
67
+ title={BARThez: a Skilled Pretrained French Sequence-to-Sequence Model},
68
+ author={Eddine, Moussa Kamal and Tixier, Antoine J-P and Vazirgiannis, Michalis},
69
+ journal={arXiv preprint arXiv:2010.12321},
70
+ year={2020}
71
+ }
72
+
73
+
74
+
75
+ ## This Dataset
76
+
77
+
78
+
79
+ ## License
80
+ CC-BY-SA-4.0