Datasets:

Modalities:
Text
Formats:
parquet
Languages:
French
ArXiv:
DOI:
Libraries:
Datasets
Dask
bourdoiscatie commited on
Commit
4d738ae
1 Parent(s): c16979f

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +278 -6
README.md CHANGED
@@ -27,8 +27,8 @@ multilinguality:
27
 
28
  # Dataset Card for Dataset of French Prompts (DFP)
29
 
30
- This dataset of prompts contains **112,194,043 rows** but for licensing reasons we can only share 106,931,863 rows. It presents data for **28 different NLP tasks**.
31
- **673 prompts** were written, including requests in imperative, tutoiement and vouvoiement form in an attempt to have as much coverage as possible of the pre-training data used by the model that will use these data and which are unknown to us.
32
 
33
  This dataset contains four columns:
34
  - inputs (string)
@@ -40,12 +40,12 @@ The `inputs` and `targets` columns follow the same format as the [xP3](https://h
40
  The `dataset` column allows the user to filter the datasets he wants to keep for his work.
41
  The `task` column allows the user to filter the tasks he wants to keep for his work.
42
 
43
- The dataset was created from 31 other datasets each with its own license. We invite you to consult them.
44
- The 673 prompts are licensed under the `cc-by-4.0` license, so you're free to apply them to your own datasets.
45
- The dataset is the concatenation of 68 prompts datasets that you can find [here](https://huggingface.co/CATIE-AQ).
46
  The nomenclature adopted for these datasets is `original dataset name` + `_fr_prompt_` + `task name`.
47
 
48
- Below, you'll find for each of the 28 tasks, the list of prompts used for each, an example of a line, the list of original datasets to which the prompts were applied and the list of datasets with prompts then created, and for each their license.
49
 
50
 
51
  <details>
@@ -2873,6 +2873,278 @@ The `universal_dependencies_fr_fqb_fr_prompt_pos`, `universal_dependencies_fr_gs
2873
  </details>
2874
 
2875
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2876
 
2877
  # Citation
2878
  Add DOI
 
27
 
28
  # Dataset Card for Dataset of French Prompts (DFP)
29
 
30
+ This dataset of prompts contains **112,194,043 rows [A ACTUALISER]** but for licensing reasons we can only share 106,931,863 rows [A ACTUALISER]. It presents data for **30 different NLP tasks**.
31
+ **724 prompts** were written, including requests in imperative, tutoiement and vouvoiement form in an attempt to have as much coverage as possible of the pre-training data used by the model that will use these data and which are unknown to us.
32
 
33
  This dataset contains four columns:
34
  - inputs (string)
 
40
  The `dataset` column allows the user to filter the datasets he wants to keep for his work.
41
  The `task` column allows the user to filter the tasks he wants to keep for his work.
42
 
43
+ The dataset was created from 34 other datasets each with its own license. We invite you to consult them.
44
+ The 724 prompts are licensed under the `cc-by-4.0` license, so you're free to apply them to your own datasets.
45
+ The dataset is the concatenation of 74 prompts datasets that you can find [here](https://huggingface.co/CATIE-AQ).
46
  The nomenclature adopted for these datasets is `original dataset name` + `_fr_prompt_` + `task name`.
47
 
48
+ Below, you'll find for each of the 30 tasks, the list of prompts used for each, an example of a line, the list of original datasets to which the prompts were applied and the list of datasets with prompts then created, and for each their license.
49
 
50
 
51
  <details>
 
2873
  </details>
2874
 
2875
 
2876
+ <details>
2877
+ <summary><h1>Data-to-text</h1></summary>
2878
+ Text generation from keywords.
2879
+
2880
+ ## 30 prompts
2881
+ <code>
2882
+ # Data-to-text (30)
2883
+
2884
+ 'Assembler les concepts suivants pour former une phrase : "'+concepts+'".', <br>
2885
+ 'Assemble les concepts suivants pour former une phrase : "'+concepts+'".', <br>
2886
+ 'Assemblez les concepts suivants pour former une phrase : "'+concepts+'".', <br>
2887
+ 'Étant donné la liste des concepts : "'+concepts+'". Générer une phrase avec tous les concepts : ', <br>
2888
+ 'Étant donné la liste des concepts : "'+concepts+'". Génère une phrase avec tous les concepts : ', <br>
2889
+ 'Étant donné la liste des concepts : "'+concepts+'". Générez une phrase avec tous les concepts : ', <br>
2890
+ 'Convertir les concepts en une phrase : "'+concepts+'".', <br>
2891
+ 'Convertis les concepts en une phrase : "'+concepts+'".', <br>
2892
+ 'Convertissez les concepts en une phrase : "'+concepts+'".', <br>
2893
+ 'Combiner tous les concepts suivants dans un texte concis et grammaticalement correct "'+concepts+'". Texte : ', <br>
2894
+ 'Combine tous les concepts suivants dans un texte concis et grammaticalement correct "'+concepts+'". Texte : ', <br>
2895
+ 'Combinez tous les concepts suivants dans un texte concis et grammaticalement correct "'+concepts+'". Texte : ', <br>
2896
+ 'Générer une phrase à partir des informations fournies ci-contre : "'+concepts+'".', <br>
2897
+ 'Génère une phrase à partir des informations fournies ci-contre : "'+concepts+'".', <br>
2898
+ 'Générez une phrase à partir des informations fournies ci-contre : "'+concepts+'".', <br>
2899
+ 'Verbaliser les concepts suivants séparés par une virgule : "'+concepts+'".', <br>
2900
+ 'Verbalise les concepts suivants séparés par une virgule : "'+concepts+'".', <br>
2901
+ 'Verbalisez les concepts suivants séparés par une virgule : "'+concepts+'".', <br>
2902
+ 'Générer un texte intégrant les concepts suivants '+concepts+'". Texte :', <br>
2903
+ 'Génère un texte intégrant les concepts suivants '+concepts+'". Texte :', <br>
2904
+ 'Générez un texte intégrant les concepts suivants '+concepts+'". Texte :', <br>
2905
+ '"'+concepts+'". Ecrire 1 à 5 phrases sur les concepts précédents.', <br>
2906
+ '"'+concepts+'". Ecris 1 à 5 phrases sur les concepts précédents.', <br>
2907
+ '"'+concepts+'". Ecrivez 1 à 5 phrases sur les concepts précédents.', <br>
2908
+ 'Rédiger un texte avec : "'+concepts+'".', <br>
2909
+ 'Rédige un texte avec : "'+concepts+'".', <br>
2910
+ 'Rédigez un texte avec : "'+concepts+'".', <br>
2911
+ 'Écrire un texte sur les concepts suivants : "'+concepts+'".', <br>
2912
+ 'Écris un texte sur les concepts suivants : "'+concepts+'".', <br>
2913
+ 'Écrivez un texte sur les concepts suivants : "'+concepts+'".',
2914
+ </code>
2915
+
2916
+ An example:
2917
+ | inputs | targets |
2918
+ | -------- | ------- |
2919
+ | Assembler les concepts suivants pour former une phrase : "Mouillabilité, Caoutchouc, Ferromagnétique, Aimantation". | Contrôle magnétique de la mouillabilité Un film de caoutchouc comportant des grains ferromagnétiques durs (avec un axe d'aimantation privilégié) est préparé avec des régions en ruban, alternées en aimantation. Si un tel film, fixé sur un support solide, est soumis à un champ magnétique tangentiel H, il doit déformer la surface libre en crêtes et vallées, et devenir de ce fait plus mouillable. |
2920
+
2921
+ ## Datasets
2922
+ ### taln-ls2n/termith-eval
2923
+ **Original**: https://huggingface.co/datasets/taln-ls2n/termith-eval
2924
+
2925
+ <details>
2926
+ <summary>Citation and License</summary>
2927
+
2928
+ #### Citation
2929
+ > > - (Boudin, 2013) Florian Boudin. 2013.
2930
+ [TALN Archives : a digital archive of French research articles in Natural Language Processing (TALN Archives : une archive numérique francophone des articles de recherche en Traitement Automatique de la Langue) [in French]][boudin-2013].
2931
+ In Proceedings of TALN 2013 (Volume 2: Short Papers), pages 507–514, Les Sables d’Olonne, France. ATALA.
2932
+ >- (Boudin and Gallina, 2021) Florian Boudin and Ygor Gallina. 2021.
2933
+ [Redefining Absent Keyphrases and their Effect on Retrieval Effectiveness][boudin-2021].
2934
+ In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 4185–4193, Online. Association for Computational Linguistics.
2935
+
2936
+ [boudin-2013]: https://aclanthology.org/F13-2001/
2937
+ [boudin-2021]: https://aclanthology.org/2021.naacl-main.330/
2938
+ #### License
2939
+ cc-by-4.0
2940
+ </details>
2941
+
2942
+ **With prompts**: https://huggingface.co/datasets/CATIE-AQ/termith-eval_fr_prompt_data_to_text
2943
+
2944
+ <details>
2945
+ <summary>Citation and License</summary>
2946
+
2947
+ #### Citation
2948
+ See the DOI at the end of this dataset card.
2949
+ #### License
2950
+ The `termith-eval_fr_prompt_data_to_text` dataset has the same license as the original dataset from which it is derived.
2951
+
2952
+
2953
+ ### taln-ls2n/taln-archives
2954
+ **Original**: https://huggingface.co/datasets/taln-ls2n/taln-archives
2955
+
2956
+ <details>
2957
+ <summary>Citation and License</summary>
2958
+
2959
+ #### Citation
2960
+ > > - (Boudin, 2013) Florian Boudin. 2013.
2961
+ [TALN Archives : a digital archive of French research articles in Natural Language Processing (TALN Archives : une archive numérique francophone des articles de recherche en Traitement Automatique de la Langue) [in French]][boudin-2013].
2962
+ In Proceedings of TALN 2013 (Volume 2: Short Papers), pages 507–514, Les Sables d’Olonne, France. ATALA.
2963
+ >- (Boudin and Gallina, 2021) Florian Boudin and Ygor Gallina. 2021.
2964
+ [Redefining Absent Keyphrases and their Effect on Retrieval Effectiveness][boudin-2021].
2965
+ In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 4185–4193, Online. Association for Computational Linguistics.
2966
+
2967
+ [boudin-2013]: https://aclanthology.org/F13-2001/
2968
+ [boudin-2021]: https://aclanthology.org/2021.naacl-main.330/
2969
+ #### License
2970
+ cc-by-4.0
2971
+ </details>
2972
+
2973
+ **With prompts**: https://huggingface.co/datasets/CATIE-AQ/taln-archives_fr_prompt_data_to_text
2974
+
2975
+ <details>
2976
+ <summary>Citation and License</summary>
2977
+
2978
+ #### Citation
2979
+ See the DOI at the end of this dataset card.
2980
+ #### License
2981
+ The `taln-archives_fr_prompt_data_to_text` dataset has the same license as the original dataset from which it is derived.
2982
+ </details>
2983
+
2984
+
2985
+ ### taln-ls2n/wikinews-fr-100
2986
+ **Original**: https://huggingface.co/datasets/taln-ls2n/wikinews-fr-100
2987
+
2988
+ <details>
2989
+ <summary>Citation and License</summary>
2990
+
2991
+ #### Citation
2992
+ > > - (Boudin, 2013) Florian Boudin. 2013.
2993
+ [TALN Archives : a digital archive of French research articles in Natural Language Processing (TALN Archives : une archive numérique francophone des articles de recherche en Traitement Automatique de la Langue) [in French]][boudin-2013].
2994
+ In Proceedings of TALN 2013 (Volume 2: Short Papers), pages 507–514, Les Sables d’Olonne, France. ATALA.
2995
+ >- (Boudin and Gallina, 2021) Florian Boudin and Ygor Gallina. 2021.
2996
+ [Redefining Absent Keyphrases and their Effect on Retrieval Effectiveness][boudin-2021].
2997
+ In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 4185–4193, Online. Association for Computational Linguistics.
2998
+
2999
+ [boudin-2013]: https://aclanthology.org/F13-2001/
3000
+ [boudin-2021]: https://aclanthology.org/2021.naacl-main.330/
3001
+ #### License
3002
+ cc-by-4.0
3003
+ </details>
3004
+
3005
+ **With prompts**: https://huggingface.co/datasets/CATIE-AQ/wikinews-fr-100_fr_prompt_data_to_text
3006
+
3007
+ <details>
3008
+ <summary>Citation and License</summary>
3009
+
3010
+ #### Citation
3011
+ See the DOI at the end of this dataset card.
3012
+ #### License
3013
+ The `wikinews-fr-100_fr_prompt_data_to_text` dataset has the same license as the original dataset from which it is derived.
3014
+ </details>
3015
+ </details>
3016
+ </details>
3017
+
3018
+
3019
+ <details>
3020
+ <summary><h1>Keywords extraction</h1></summary>
3021
+
3022
+ ## 21 prompts
3023
+ <code>
3024
+ 'Extraire les mots clés importants du texte suivant : '+text, <br>
3025
+ 'Extrais les mots clés importants du texte suivant : '+text, <br>
3026
+ 'Extrayez les mots clés importants du texte suivant : '+text, <br>
3027
+ 'Isoler les mots clés importants du texte suivant : '+text, <br>
3028
+ 'Isole les mots clés importants du texte suivant : '+text, <br>
3029
+ 'Isolez les mots clés importants du texte suivant : '+text, <br>
3030
+ 'Dégager des mots clés dans le texte : '+text, <br>
3031
+ 'Dégage des mots clés dans le texte : '+text, <br>
3032
+ 'Dégagez des mots clés dans le texte : '+text, <br>
3033
+ 'Générer des mots clés issus du texte suivant : '+text, <br>
3034
+ 'Génère des mots clés issus du texte suivant : '+text, <br>
3035
+ 'Générez des mots clés issus du texte suivant : '+text, <br>
3036
+ 'Trouver les mots clés du texte : '+text, <br>
3037
+ 'Trouve les mots clés du texte : '+text, <br>
3038
+ 'Trouvez les mots clés du texte : '+text, <br>
3039
+ 'Repérer les mots clés importants présents dans le texte suivant : '+text, <br>
3040
+ 'Repère les mots clés importants présents dans le texte suivant : '+text, <br>
3041
+ 'Repérez les mots clés importants présents dans le texte suivant : '+text, <br>
3042
+ 'Indiquer les mots clés du texte : '+text, <br>
3043
+ 'Indiquer les mots clés du texte : '+text, <br>
3044
+ 'Indiquer les mots clés du texte : '+text
3045
+ </code>
3046
+
3047
+ An example:
3048
+ | inputs | targets |
3049
+ | -------- | ------- |
3050
+ | "Extraire les mots clés importants du texte suivant : Contrôle magnétique de la mouillabilité Un film de caoutchouc comportant des grains ferromagnétiques durs (avec un axe d'aimantation privilégié) est préparé avec des régions en ruban, alternées en aimantation. Si un tel film, fixé sur un support solide, est soumis à un champ magnétique tangentiel H, il doit déformer la surface libre en crêtes et vallées, et devenir de ce fait plus mouillable." | "Mouillabilité, Caoutchouc, Ferromagnétique, Aimantation". |
3051
+
3052
+
3053
+ ## Datasets
3054
+ ### taln-ls2n/termith-eval
3055
+ **Original**: https://huggingface.co/datasets/taln-ls2n/termith-eval
3056
+
3057
+ <details>
3058
+ <summary>Citation and License</summary>
3059
+
3060
+ #### Citation
3061
+ > > - (Boudin, 2013) Florian Boudin. 2013.
3062
+ [TALN Archives : a digital archive of French research articles in Natural Language Processing (TALN Archives : une archive numérique francophone des articles de recherche en Traitement Automatique de la Langue) [in French]][boudin-2013].
3063
+ In Proceedings of TALN 2013 (Volume 2: Short Papers), pages 507–514, Les Sables d’Olonne, France. ATALA.
3064
+ >- (Boudin and Gallina, 2021) Florian Boudin and Ygor Gallina. 2021.
3065
+ [Redefining Absent Keyphrases and their Effect on Retrieval Effectiveness][boudin-2021].
3066
+ In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 4185–4193, Online. Association for Computational Linguistics.
3067
+
3068
+ [boudin-2013]: https://aclanthology.org/F13-2001/
3069
+ [boudin-2021]: https://aclanthology.org/2021.naacl-main.330/
3070
+ #### License
3071
+ cc-by-4.0
3072
+ </details>
3073
+
3074
+ **With prompts**: https://huggingface.co/datasets/CATIE-AQ/termith-eval_fr_prompt_keywords_extraction
3075
+
3076
+ <details>
3077
+ <summary>Citation and License</summary>
3078
+
3079
+ #### Citation
3080
+ See the DOI at the end of this dataset card.
3081
+ #### License
3082
+ The `termith-eval_fr_prompt_keywords_extraction` dataset has the same license as the original dataset from which it is derived.
3083
+
3084
+
3085
+ ### taln-ls2n/taln-archives
3086
+ **Original**: https://huggingface.co/datasets/taln-ls2n/taln-archives
3087
+
3088
+ <details>
3089
+ <summary>Citation and License</summary>
3090
+
3091
+ #### Citation
3092
+ > > - (Boudin, 2013) Florian Boudin. 2013.
3093
+ [TALN Archives : a digital archive of French research articles in Natural Language Processing (TALN Archives : une archive numérique francophone des articles de recherche en Traitement Automatique de la Langue) [in French]][boudin-2013].
3094
+ In Proceedings of TALN 2013 (Volume 2: Short Papers), pages 507–514, Les Sables d’Olonne, France. ATALA.
3095
+ >- (Boudin and Gallina, 2021) Florian Boudin and Ygor Gallina. 2021.
3096
+ [Redefining Absent Keyphrases and their Effect on Retrieval Effectiveness][boudin-2021].
3097
+ In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 4185–4193, Online. Association for Computational Linguistics.
3098
+
3099
+ [boudin-2013]: https://aclanthology.org/F13-2001/
3100
+ [boudin-2021]: https://aclanthology.org/2021.naacl-main.330/
3101
+ #### License
3102
+ cc-by-4.0
3103
+ </details>
3104
+
3105
+ **With prompts**: https://huggingface.co/datasets/CATIE-AQ/taln-archives_fr_prompt_keywords_extraction
3106
+
3107
+ <details>
3108
+ <summary>Citation and License</summary>
3109
+
3110
+ #### Citation
3111
+ See the DOI at the end of this dataset card.
3112
+ #### License
3113
+ The `taln-archives_fr_prompt_keywords_extraction` dataset has the same license as the original dataset from which it is derived.
3114
+ </details>
3115
+
3116
+
3117
+ ### taln-ls2n/wikinews-fr-100
3118
+ **Original**: https://huggingface.co/datasets/taln-ls2n/wikinews-fr-100
3119
+
3120
+ <details>
3121
+ <summary>Citation and License</summary>
3122
+
3123
+ #### Citation
3124
+ > > - (Boudin, 2013) Florian Boudin. 2013.
3125
+ [TALN Archives : a digital archive of French research articles in Natural Language Processing (TALN Archives : une archive numérique francophone des articles de recherche en Traitement Automatique de la Langue) [in French]][boudin-2013].
3126
+ In Proceedings of TALN 2013 (Volume 2: Short Papers), pages 507–514, Les Sables d’Olonne, France. ATALA.
3127
+ >- (Boudin and Gallina, 2021) Florian Boudin and Ygor Gallina. 2021.
3128
+ [Redefining Absent Keyphrases and their Effect on Retrieval Effectiveness][boudin-2021].
3129
+ In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 4185–4193, Online. Association for Computational Linguistics.
3130
+
3131
+ [boudin-2013]: https://aclanthology.org/F13-2001/
3132
+ [boudin-2021]: https://aclanthology.org/2021.naacl-main.330/
3133
+ #### License
3134
+ cc-by-4.0
3135
+ </details>
3136
+
3137
+ **With prompts**: https://huggingface.co/datasets/CATIE-AQ/wikinews-fr-100_fr_prompt_keywords_extraction
3138
+
3139
+ <details>
3140
+ <summary>Citation and License</summary>
3141
+
3142
+ #### Citation
3143
+ See the DOI at the end of this dataset card.
3144
+ #### License
3145
+ The `wikinews-fr-100_fr_prompt_keywords_extraction` dataset has the same license as the original dataset from which it is derived.
3146
+ </details>
3147
+ </details>
3148
 
3149
  # Citation
3150
  Add DOI