Hugging Face's logo

qanastek
/
pos-french-camembert-flair

Token Classification
Flair
PyTorch
French
sequence-tagger-model
Model card Files Community
qanastek commited on
Commit
44e1596
•
1 Parent(s): 74192fe

Update model

Browse files
Files changed (1) hide show
  1. README.md +64 -62
README.md CHANGED
@@ -13,9 +13,9 @@ widget:
13
  # POET: A French Extended Part-of-Speech Tagger
14
 
15
  - Corpora: [ANTILLES](https://github.com/qanastek/ANTILLES)
16
- - Embeddings: [FastText](https://fasttext.cc/)
17
  - Sequence Labelling: [Bi-LSTM-CRF](https://arxiv.org/abs/1011.4088)
18
- - Number of Epochs: 115
19
 
20
  **People Involved**
21
 
@@ -143,76 +143,78 @@ The test corpora used for this evaluation is available on [Github](https://githu
143
 
144
  ```plain
145
  Results:
146
- - F-score (micro): 0.952
147
- - F-score (macro): 0.8644
148
- - Accuracy (incl. no class): 0.952
149
 
150
  By class:
151
  precision recall f1-score support
152
- PPER1S 0.9767 1.0000 0.9882 42
153
- VERB 0.9823 0.9537 0.9678 583
154
- COSUB 0.9344 0.8906 0.9120 128
155
- PUNCT 0.9878 0.9688 0.9782 833
156
- PREP 0.9767 0.9879 0.9822 1483
157
- PDEMMS 0.9583 0.9200 0.9388 75
158
- COCO 0.9839 1.0000 0.9919 245
159
- DET 0.9679 0.9814 0.9746 645
160
- NMP 0.9521 0.9115 0.9313 305
161
- ADJMP 0.8352 0.9268 0.8786 82
162
- PREL 0.9324 0.9857 0.9583 70
163
- PREFP 0.9767 0.9545 0.9655 44
164
- AUX 0.9537 0.9859 0.9695 355
165
- ADV 0.9440 0.9365 0.9402 504
166
- VPPMP 0.8667 1.0000 0.9286 26
167
- DINTMS 0.9919 1.0000 0.9959 122
168
- ADJMS 0.9020 0.9057 0.9039 244
169
- NMS 0.9226 0.9336 0.9281 753
170
- NFS 0.9347 0.9714 0.9527 560
171
- YPFOR 0.9806 1.0000 0.9902 353
172
- PINDMS 1.0000 0.9091 0.9524 44
173
- NOUN 0.8400 0.5385 0.6562 39
174
- PROPN 0.8605 0.8278 0.8439 395
175
- DETMS 0.9972 0.9972 0.9972 362
176
- PPER3MS 0.9341 0.9770 0.9551 87
177
- VPPMS 0.8994 0.9682 0.9325 157
178
  DETFS 1.0000 1.0000 1.0000 240
179
- ADJFS 0.9266 0.9011 0.9136 182
180
- ADJFP 0.9726 0.9342 0.9530 76
181
- NFP 0.9463 0.9749 0.9604 199
182
- VPPFS 0.8000 0.9000 0.8471 40
183
- CHIF 0.9543 0.9414 0.9478 222
184
- XFAMIL 0.9346 0.8696 0.9009 115
185
- PPER3MP 0.9474 0.9000 0.9231 20
186
- PPOBJMS 0.8800 0.9362 0.9072 47
187
- PREF 0.8889 0.9231 0.9057 52
188
- PPOBJMP 1.0000 0.6000 0.7500 10
189
- SYM 0.9706 0.8684 0.9167 38
190
- DINTFS 0.9683 1.0000 0.9839 61
191
- PDEMFS 1.0000 0.8966 0.9455 29
192
- PPER3FS 1.0000 0.9444 0.9714 18
193
- VPPFP 0.9500 1.0000 0.9744 19
194
- PRON 0.9200 0.7419 0.8214 31
195
- PPOBJFS 0.8333 0.8333 0.8333 6
196
- PART 0.8000 1.0000 0.8889 4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
197
  PPER3FP 1.0000 1.0000 1.0000 2
198
- MOTINC 0.3571 0.3333 0.3448 15
199
- PDEMMP 1.0000 0.6667 0.8000 3
200
- INTJ 0.4000 0.6667 0.5000 6
201
- PREFS 1.0000 0.5000 0.6667 10
202
- ADJ 0.7917 0.8636 0.8261 22
203
- PINDMP 0.0000 0.0000 0.0000 1
204
- PINDFS 1.0000 1.0000 1.0000 1
205
  NUM 1.0000 0.3333 0.5000 3
206
  PPER2S 1.0000 1.0000 1.0000 2
207
- PPOBJFP 1.0000 0.5000 0.6667 2
208
- PDEMFP 1.0000 0.6667 0.8000 3
209
- X 0.0000 0.0000 0.0000 1
210
  PRELMS 1.0000 1.0000 1.0000 2
 
 
 
211
  PINDFP 1.0000 1.0000 1.0000 1
212
 
213
- accuracy 0.9520 10019
214
- macro avg 0.8956 0.8521 0.8644 10019
215
- weighted avg 0.9524 0.9520 0.9515 10019
 
216
  ```
217
 
218
  ## BibTeX Citations
 
13
  # POET: A French Extended Part-of-Speech Tagger
14
 
15
  - Corpora: [ANTILLES](https://github.com/qanastek/ANTILLES)
16
+ - Embeddings: [Flair](https://aclanthology.org/C18-1139.pdf) & [CamemBERT](https://arxiv.org/abs/1911.03894)
17
  - Sequence Labelling: [Bi-LSTM-CRF](https://arxiv.org/abs/1011.4088)
18
+ - Number of Epochs: 50
19
 
20
  **People Involved**
21
 
 
143
 
144
  ```plain
145
  Results:
146
+ - F-score (micro) 0.9797
147
+ - F-score (macro) 0.9178
148
+ - Accuracy 0.9797
149
 
150
  By class:
151
  precision recall f1-score support
152
+
153
+ PREP 0.9966 0.9987 0.9976 1483
154
+ PUNCT 1.0000 1.0000 1.0000 833
155
+ NMS 0.9634 0.9801 0.9717 753
156
+ DET 0.9923 0.9984 0.9954 645
157
+ VERB 0.9913 0.9811 0.9862 583
158
+ NFS 0.9667 0.9839 0.9752 560
159
+ ADV 0.9940 0.9821 0.9880 504
160
+ PROPN 0.9541 0.8937 0.9229 395
161
+ DETMS 1.0000 1.0000 1.0000 362
162
+ AUX 0.9860 0.9915 0.9888 355
163
+ YPFOR 1.0000 1.0000 1.0000 353
164
+ NMP 0.9666 0.9475 0.9570 305
165
+ COCO 0.9959 1.0000 0.9980 245
166
+ ADJMS 0.9463 0.9385 0.9424 244
 
 
 
 
 
 
 
 
 
 
 
167
  DETFS 1.0000 1.0000 1.0000 240
168
+ CHIF 0.9648 0.9865 0.9755 222
169
+ NFP 0.9515 0.9849 0.9679 199
170
+ ADJFS 0.9657 0.9286 0.9468 182
171
+ VPPMS 0.9387 0.9745 0.9563 157
172
+ COSUB 1.0000 0.9844 0.9921 128
173
+ DINTMS 0.9918 0.9918 0.9918 122
174
+ XFAMIL 0.9298 0.9217 0.9258 115
175
+ PPER3MS 1.0000 1.0000 1.0000 87
176
+ ADJMP 0.9294 0.9634 0.9461 82
177
+ PDEMMS 1.0000 1.0000 1.0000 75
178
+ ADJFP 0.9861 0.9342 0.9595 76
179
+ PREL 0.9859 1.0000 0.9929 70
180
+ DINTFS 0.9839 1.0000 0.9919 61
181
+ PREF 1.0000 1.0000 1.0000 52
182
+ PPOBJMS 0.9565 0.9362 0.9462 47
183
+ PREFP 0.9778 1.0000 0.9888 44
184
+ PINDMS 1.0000 0.9773 0.9885 44
185
+ VPPFS 0.8298 0.9750 0.8966 40
186
+ PPER1S 1.0000 1.0000 1.0000 42
187
+ SYM 1.0000 0.9474 0.9730 38
188
+ NOUN 0.8824 0.7692 0.8219 39
189
+ PRON 1.0000 0.9677 0.9836 31
190
+ PDEMFS 1.0000 1.0000 1.0000 29
191
+ VPPMP 0.9286 1.0000 0.9630 26
192
+ ADJ 0.9524 0.9091 0.9302 22
193
+ PPER3MP 1.0000 1.0000 1.0000 20
194
+ VPPFP 1.0000 1.0000 1.0000 19
195
+ PPER3FS 1.0000 1.0000 1.0000 18
196
+ MOTINC 0.3333 0.4000 0.3636 15
197
+ PREFS 1.0000 1.0000 1.0000 10
198
+ PPOBJMP 1.0000 0.8000 0.8889 10
199
+ PPOBJFS 0.6250 0.8333 0.7143 6
200
+ INTJ 0.5000 0.6667 0.5714 6
201
+ PART 1.0000 1.0000 1.0000 4
202
+ PDEMMP 1.0000 1.0000 1.0000 3
203
+ PDEMFP 1.0000 1.0000 1.0000 3
204
  PPER3FP 1.0000 1.0000 1.0000 2
 
 
 
 
 
 
 
205
  NUM 1.0000 0.3333 0.5000 3
206
  PPER2S 1.0000 1.0000 1.0000 2
207
+ PPOBJFP 0.5000 0.5000 0.5000 2
 
 
208
  PRELMS 1.0000 1.0000 1.0000 2
209
+ PINDFS 0.5000 1.0000 0.6667 1
210
+ PINDMP 1.0000 1.0000 1.0000 1
211
+ X 0.0000 0.0000 0.0000 1
212
  PINDFP 1.0000 1.0000 1.0000 1
213
 
214
+ micro avg 0.9797 0.9797 0.9797 10019
215
+ macro avg 0.9228 0.9230 0.9178 10019
216
+ weighted avg 0.9802 0.9797 0.9798 10019
217
+ samples avg 0.9797 0.9797 0.9797 10019
218
  ```
219
 
220
  ## BibTeX Citations