Hugging Face's logo

AntoineBourgois
/
propp-fr_NER_camembert-large_PER

Token Classification
French
NER
camembert
literary-texts
nested-entities
propp-fr
Model card Files
xet
 Community
AntoineBourgois commited on
Commit
8babf6c
·
verified ·
1 Parent(s): b628522

Upload 3 files

Browse files
Files changed (3) hide show
  1. JCLS_model_card.md +115 -0
  2. README.md +115 -0
  3. final_model.pkl +3 -0
JCLS_model_card.md ADDED
@@ -0,0 +1,115 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ ---
3
+ language: fr
4
+ tags:
5
+ - NER
6
+ - camembert
7
+ - literary-texts
8
+ - nested-entities
9
+ - BookNLP-fr
10
+ license: apache-2.0
11
+ metrics:
12
+ - f1
13
+ - precision
14
+ - recall
15
+ base_model:
16
+ - almanach/camembert-large
17
+ pipeline_tag: token-classification
18
+ ---
19
+
20
+ ## INTRODUCTION:
21
+ This model, developed as part of the [BookNLP-fr project](https://github.com/lattice-8094/fr-litbank), is a **NER model** built on top of [camembert-large](https://huggingface.co/almanach/camembert-large) embeddings, trained to predict nested entities in french, specifically for literary texts.
22
+
23
+ The predicted entities are:
24
+ - mentions of characters (PER): pronouns (je, tu, il, ...), possessive pronouns (mon, ton, son, ...), common nouns (le capitaine, la princesse, ...) and proper nouns (Indiana Delmare, Honoré de Pardaillan, ...)
25
+ - facilities (FAC): chatêau, sentier, chambre, couloir, ...
26
+ - time (TIME): le règne de Louis XIV, ce matin, en juillet, ...
27
+ - geo-political entities (GPE): Montrouge, France, le petit hameau, ...
28
+ - locations (LOC): le sud, Mars, l'océan, le bois, ...
29
+ - vehicles (VEH): avion, voitures, calèche, vélos, ...
30
+
31
+ ## MODEL PERFORMANCES (LOOCV):
32
+ | NER_tag | precision | recall | f1_score | support | support % |
33
+ |-----------|-------------|----------|------------|-----------|-------------|
34
+ | PER | 91.54% | 95.35% | 93.40% | 4,061 | 100.00% |
35
+ | micro_avg | 91.54% | 95.35% | 93.40% | 4,061 | 100.00% |
36
+ | macro_avg | 91.54% | 95.35% | 93.40% | 4,061 | 100.00% |
37
+
38
+ ## TRAINING PARAMETERS:
39
+ - Entities types: ['PER']
40
+ - Tagging scheme: BIOES
41
+ - Nested entities levels: [0, 1]
42
+ - Split strategy: Leave-one-out cross-validation (31 files)
43
+ - Train/Validation split: 0.85 / 0.15
44
+ - Batch size: 16
45
+ - Initial learning rate: 0.00014
46
+
47
+ ## MODEL ARCHITECTURE:
48
+ Model Input: Maximum context camembert-large embeddings (1024 dimensions)
49
+
50
+ - Locked Dropout: 0.5
51
+
52
+ - Projection layer:
53
+ - layer type: highway layer
54
+ - input: 1024 dimensions
55
+ - output: 2048 dimensions
56
+
57
+ - BiLSTM layer:
58
+ - input: 2048 dimensions
59
+ - output: 256 dimensions (hidden state)
60
+
61
+ - Linear layer:
62
+ - input: 256 dimensions
63
+ - output: 5 dimensions (predicted labels with BIOES tagging scheme)
64
+
65
+ - CRF layer
66
+
67
+ Model Output: BIOES labels sequence
68
+
69
+ ## HOW TO USE:
70
+ *** IN CONSTRUCTION ***
71
+
72
+ ## TRAINING CORPUS:
73
+ | | Document | Tokens Count | Is included in model eval |
74
+ |----|----------------------------------------------------------------|----------------|-----------------------------------|
75
+ | 0 | 1731_Prévost-Antoine-François_Manon-Lescaut_PER-ONLY | 71,219 tokens | False |
76
+ | 1 | 1832_Sand-George_Indiana_PER-ONLY | 112,221 tokens | False |
77
+ | 2 | 1836_Gautier-Theophile_La-morte-amoureuse | 14,293 tokens | False |
78
+ | 3 | 1840_Sand-George_Pauline | 12,407 tokens | False |
79
+ | 4 | 1842_Balzac-Honore-de_La-Maison-du-chat-qui-pelote | 24,776 tokens | False |
80
+ | 5 | 1844_Balzac-Honore-de_La-Maison-Nucingen | 30,034 tokens | False |
81
+ | 6 | 1844_Balzac-Honore-de_Sarrasine | 15,408 tokens | False |
82
+ | 7 | 1856_Cousin-Victor_Madame-de-Hautefort | 11,768 tokens | False |
83
+ | 8 | 1863_Gautier-Theophile_Le-capitaine-Fracasse | 11,855 tokens | False |
84
+ | 9 | 1873_Zola-Emile_Le-ventre-de-Paris | 12,617 tokens | False |
85
+ | 10 | 1881_Flaubert-Gustave_Bouvard-et-Pecuchet | 12,320 tokens | False |
86
+ | 11 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-1_1-MADEMOISELLE-FIFI | 5,449 tokens | True |
87
+ | 12 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-1_2-MADAME-BAPTISTE | 2,579 tokens | True |
88
+ | 13 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-1_3-LA-ROUILLE | 2,949 tokens | True |
89
+ | 14 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-2_1-MARROCA | 4,081 tokens | False |
90
+ | 15 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-2_2-LA-BUCHE | 2,267 tokens | False |
91
+ | 16 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-2_3-LA-RELIQUE | 2,042 tokens | False |
92
+ | 17 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-3_1-FOU | 1,906 tokens | False |
93
+ | 18 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-3_2-REVEIL | 2,160 tokens | False |
94
+ | 19 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-3_3-UNE-RUSE | 2,470 tokens | False |
95
+ | 20 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-3_4-A-CHEVAL | 2,880 tokens | False |
96
+ | 21 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-3_5-UN-REVEILLON | 2,365 tokens | False |
97
+ | 22 | 1901_Lucie-Achard_Rosalie-de-Constant-sa-famille-et-ses-amis | 12,789 tokens | False |
98
+ | 23 | 1903_Conan-Laure_Elisabeth_Seton | 13,054 tokens | False |
99
+ | 24 | 1904_Rolland-Romain_Jean-Christophe_Tome-I-L-aube | 10,982 tokens | True |
100
+ | 25 | 1904_Rolland-Romain_Jean-Christophe_Tome-II-Le-matin | 10,305 tokens | False |
101
+ | 26 | 1917_Adèle-Bourgeois_Némoville | 12,478 tokens | False |
102
+ | 27 | 1923_Delly_Dans-les-ruines | 95,617 tokens | False |
103
+ | 28 | 1923_Radiguet-Raymond_Le-diable-au-corps | 14,860 tokens | False |
104
+ | 29 | 1926_Audoux-Marguerite_De-la-ville-au-moulin | 12,092 tokens | True |
105
+ | 30 | 1937_Audoux-Marguerite_Douce-Lumiere | 12,348 tokens | False |
106
+ | 31 | TOTAL | 554,591 tokens | 5 files used for cross-validation |
107
+
108
+ ## PREDICTIONS CONFUSION MATRIX:
109
+ | Gold Labels | PER | O | support |
110
+ |---------------|-------|-----|-----------|
111
+ | PER | 3,872 | 189 | 4,061 |
112
+ | O | 348 | 0 | 348 |
113
+
114
+ ## CONTACT:
115
+ mail: antoine [dot] bourgois [at] protonmail [dot] com
README.md ADDED
@@ -0,0 +1,115 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ ---
3
+ language: fr
4
+ tags:
5
+ - NER
6
+ - camembert
7
+ - literary-texts
8
+ - nested-entities
9
+ - BookNLP-fr
10
+ license: apache-2.0
11
+ metrics:
12
+ - f1
13
+ - precision
14
+ - recall
15
+ base_model:
16
+ - almanach/camembert-large
17
+ pipeline_tag: token-classification
18
+ ---
19
+
20
+ ## INTRODUCTION:
21
+ This model, developed as part of the [BookNLP-fr project](https://github.com/lattice-8094/fr-litbank), is a **NER model** built on top of [camembert-large](https://huggingface.co/almanach/camembert-large) embeddings, trained to predict nested entities in french, specifically for literary texts.
22
+
23
+ The predicted entities are:
24
+ - mentions of characters (PER): pronouns (je, tu, il, ...), possessive pronouns (mon, ton, son, ...), common nouns (le capitaine, la princesse, ...) and proper nouns (Indiana Delmare, Honoré de Pardaillan, ...)
25
+ - facilities (FAC): chatêau, sentier, chambre, couloir, ...
26
+ - time (TIME): le règne de Louis XIV, ce matin, en juillet, ...
27
+ - geo-political entities (GPE): Montrouge, France, le petit hameau, ...
28
+ - locations (LOC): le sud, Mars, l'océan, le bois, ...
29
+ - vehicles (VEH): avion, voitures, calèche, vélos, ...
30
+
31
+ ## MODEL PERFORMANCES (LOOCV):
32
+ | NER_tag | precision | recall | f1_score | support | support % |
33
+ |-----------|-------------|----------|------------|-----------|-------------|
34
+ | PER | 93.53% | 94.56% | 94.04% | 71,106 | 100.00% |
35
+ | micro_avg | 93.53% | 94.56% | 94.04% | 71,106 | 100.00% |
36
+ | macro_avg | 93.53% | 94.56% | 94.04% | 71,106 | 100.00% |
37
+
38
+ ## TRAINING PARAMETERS:
39
+ - Entities types: ['PER']
40
+ - Tagging scheme: BIOES
41
+ - Nested entities levels: [0, 1]
42
+ - Split strategy: Leave-one-out cross-validation (31 files)
43
+ - Train/Validation split: 0.85 / 0.15
44
+ - Batch size: 16
45
+ - Initial learning rate: 0.00014
46
+
47
+ ## MODEL ARCHITECTURE:
48
+ Model Input: Maximum context camembert-large embeddings (1024 dimensions)
49
+
50
+ - Locked Dropout: 0.5
51
+
52
+ - Projection layer:
53
+ - layer type: highway layer
54
+ - input: 1024 dimensions
55
+ - output: 2048 dimensions
56
+
57
+ - BiLSTM layer:
58
+ - input: 2048 dimensions
59
+ - output: 256 dimensions (hidden state)
60
+
61
+ - Linear layer:
62
+ - input: 256 dimensions
63
+ - output: 5 dimensions (predicted labels with BIOES tagging scheme)
64
+
65
+ - CRF layer
66
+
67
+ Model Output: BIOES labels sequence
68
+
69
+ ## HOW TO USE:
70
+ *** IN CONSTRUCTION ***
71
+
72
+ ## TRAINING CORPUS:
73
+ | | Document | Tokens Count | Is included in model eval |
74
+ |----|----------------------------------------------------------------|----------------|------------------------------------|
75
+ | 0 | 1731_Prévost-Antoine-François_Manon-Lescaut_PER-ONLY | 71,219 tokens | True |
76
+ | 1 | 1832_Sand-George_Indiana_PER-ONLY | 112,221 tokens | True |
77
+ | 2 | 1836_Gautier-Theophile_La-morte-amoureuse | 14,293 tokens | True |
78
+ | 3 | 1840_Sand-George_Pauline | 12,407 tokens | True |
79
+ | 4 | 1842_Balzac-Honore-de_La-Maison-du-chat-qui-pelote | 24,776 tokens | True |
80
+ | 5 | 1844_Balzac-Honore-de_La-Maison-Nucingen | 30,034 tokens | True |
81
+ | 6 | 1844_Balzac-Honore-de_Sarrasine | 15,408 tokens | True |
82
+ | 7 | 1856_Cousin-Victor_Madame-de-Hautefort | 11,768 tokens | True |
83
+ | 8 | 1863_Gautier-Theophile_Le-capitaine-Fracasse | 11,855 tokens | True |
84
+ | 9 | 1873_Zola-Emile_Le-ventre-de-Paris | 12,617 tokens | True |
85
+ | 10 | 1881_Flaubert-Gustave_Bouvard-et-Pecuchet | 12,320 tokens | True |
86
+ | 11 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-1_1-MADEMOISELLE-FIFI | 5,449 tokens | True |
87
+ | 12 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-1_2-MADAME-BAPTISTE | 2,579 tokens | True |
88
+ | 13 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-1_3-LA-ROUILLE | 2,949 tokens | True |
89
+ | 14 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-2_1-MARROCA | 4,081 tokens | True |
90
+ | 15 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-2_2-LA-BUCHE | 2,267 tokens | True |
91
+ | 16 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-2_3-LA-RELIQUE | 2,042 tokens | True |
92
+ | 17 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-3_1-FOU | 1,906 tokens | True |
93
+ | 18 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-3_2-REVEIL | 2,160 tokens | True |
94
+ | 19 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-3_3-UNE-RUSE | 2,470 tokens | True |
95
+ | 20 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-3_4-A-CHEVAL | 2,880 tokens | True |
96
+ | 21 | 1882_Guy-de-Maupassant_Mademoiselle-Fifi-3_5-UN-REVEILLON | 2,365 tokens | True |
97
+ | 22 | 1901_Lucie-Achard_Rosalie-de-Constant-sa-famille-et-ses-amis | 12,789 tokens | True |
98
+ | 23 | 1903_Conan-Laure_Elisabeth_Seton | 13,054 tokens | True |
99
+ | 24 | 1904_Rolland-Romain_Jean-Christophe_Tome-I-L-aube | 10,982 tokens | True |
100
+ | 25 | 1904_Rolland-Romain_Jean-Christophe_Tome-II-Le-matin | 10,305 tokens | True |
101
+ | 26 | 1917_Adèle-Bourgeois_Némoville | 12,478 tokens | True |
102
+ | 27 | 1923_Delly_Dans-les-ruines | 95,617 tokens | True |
103
+ | 28 | 1923_Radiguet-Raymond_Le-diable-au-corps | 14,860 tokens | True |
104
+ | 29 | 1926_Audoux-Marguerite_De-la-ville-au-moulin | 12,092 tokens | True |
105
+ | 30 | 1937_Audoux-Marguerite_Douce-Lumiere | 12,348 tokens | True |
106
+ | 31 | TOTAL | 554,591 tokens | 31 files used for cross-validation |
107
+
108
+ ## PREDICTIONS CONFUSION MATRIX:
109
+ | Gold Labels | PER | O | support |
110
+ |---------------|--------|-------|-----------|
111
+ | PER | 67,239 | 3,867 | 71,106 |
112
+ | O | 4,487 | 0 | 4,487 |
113
+
114
+ ## CONTACT:
115
+ mail: antoine [dot] bourgois [at] protonmail [dot] com
final_model.pkl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:42c6586920956b6fff84a52e91a9b9f23916f0632acce4a96bcbed4b5f240cf6
3
+ size 386227046