jeanpoll commited on
Commit
5867026
1 Parent(s): 6de7397

New improved model trained on full dataset

Browse files
Files changed (2) hide show
  1. README.md +20 -17
  2. pytorch_model.bin +2 -2
README.md CHANGED
@@ -3,7 +3,7 @@ language: fr
3
  datasets:
4
  - Jean-Baptiste/wikiner_fr
5
  widget:
6
- - text: "Je m'appelle Jean-Baptiste et je vis à Paris"
7
  ---
8
 
9
  # camembert-ner: model fine-tuned from camemBERT for NER task.
@@ -11,7 +11,9 @@ widget:
11
  ## Introduction
12
 
13
  [camembert-ner] is a NER model that was fine-tuned from camemBERT on wikiner-fr dataset.
14
- Model was trained on subset of wikiner-fr dataset (~36 000 sentences)
 
 
15
 
16
 
17
  ## How to use camembert-ner with HuggingFace
@@ -81,24 +83,25 @@ nlp("Apple est créée le 1er avril 1976 dans le garage de la maison d'enfance d
81
 
82
  Global
83
  ```
84
- 'precision': 0.8830965723967158
85
- 'recall': 0.8915789473684211
86
- 'f1': 0.8873174883781837
87
  ```
88
 
89
  By entity
90
  ```
91
- 'LOC': {'precision': 0.8701754385964913,
92
- 'recall': 0.8878281622911695,
93
- 'f1': 0.8789131718842291},
94
- 'MISC': {'precision': 0.831053901850362,
95
- 'recall': 0.815955766192733,
96
- 'f1': 0.823435631725787},
97
- 'ORG': {'precision': 0.8620199146514936,
98
- 'recall': 0.8335625859697386,
99
- 'f1': 0.8475524475524475},
100
- 'PER': {'precision': 0.9367143476376246,
101
- 'recall': 0.9583148558758315,
102
- 'f1': 0.947391494958}
 
103
  ```
104
 
3
  datasets:
4
  - Jean-Baptiste/wikiner_fr
5
  widget:
6
+ - text: "Je m'appelle jean-baptiste et je vis à montréal"
7
  ---
8
 
9
  # camembert-ner: model fine-tuned from camemBERT for NER task.
11
  ## Introduction
12
 
13
  [camembert-ner] is a NER model that was fine-tuned from camemBERT on wikiner-fr dataset.
14
+ Model was trained on wikiner-fr dataset (~170 634 sentences).
15
+ Model was validated on emails/chat data and surperformed other models on this type of data specifically.
16
+ In particular the model seems to work better on entity that don't start with an upper case.
17
 
18
 
19
  ## How to use camembert-ner with HuggingFace
83
 
84
  Global
85
  ```
86
+ 'precision': 0.8859
87
+ 'recall': 0.8971
88
+ 'f1': 0.8914
89
  ```
90
 
91
  By entity
92
  ```
93
+ 'LOC': {'precision': 0.8905576596578294,
94
+ 'recall': 0.900554675118859,
95
+ 'f1': 0.8955282684352223},
96
+ 'MISC': {'precision': 0.8175627240143369,
97
+ 'recall': 0.8117437722419929,
98
+ 'f1': 0.8146428571428571},
99
+ 'ORG': {'precision': 0.8099480326651819,
100
+ 'recall': 0.8265151515151515,
101
+ 'f1': 0.8181477315335584},
102
+ 'PER': {'precision': 0.9372509960159362,
103
+ 'recall': 0.959812321501428,
104
+ 'f1': 0.9483975005039308}
105
+
106
  ```
107
 
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:55e4e2730fadcb7f4905c10d6f39abc0fe59ead7f919e5fde845f39144cc430b
3
- size 440227439
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c9f586c5bc5943992fa49fe0c0c390dace2a48288d1cec0680cd96fcd17ed037
3
+ size 440227047