abbasmahmudiai
commited on
Commit
•
0bd1a97
1
Parent(s):
3060464
Update README.md
Browse files
README.md
CHANGED
@@ -6,4 +6,41 @@ language:
|
|
6 |
pipeline_tag: token-classification
|
7 |
---
|
8 |
Named entity recognition On Persian dataset
|
|
|
|
|
|
|
9 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6 |
pipeline_tag: token-classification
|
7 |
---
|
8 |
Named entity recognition On Persian dataset
|
9 |
+
traindataset=20484 persian sentense
|
10 |
+
valdataset=2561
|
11 |
+
AutoTokenizer=HooshvareLab/bert-fa-base-uncased
|
12 |
|
13 |
+
ner_tags=
|
14 |
+
['O', 'B-pro',
|
15 |
+
'I-pro',
|
16 |
+
'B-pers',
|
17 |
+
'I-pers',
|
18 |
+
'B-org',
|
19 |
+
'I-org',
|
20 |
+
'B-loc',
|
21 |
+
'I-loc',
|
22 |
+
'B-fac',
|
23 |
+
'I-fac',
|
24 |
+
'B-event',
|
25 |
+
'I-event']
|
26 |
+
|
27 |
+
training_args=
|
28 |
+
learning_rate=2e-5,
|
29 |
+
per_device_train_batch_size=16,
|
30 |
+
per_device_eval_batch_size=16,
|
31 |
+
num_train_epochs=4,
|
32 |
+
weight_decay=0.01
|
33 |
+
|
34 |
+
Training Loss=0.001000
|
35 |
+
|
36 |
+
sample1:
|
37 |
+
'entity': 'B-loc',
|
38 |
+
'score': 0.9998902,
|
39 |
+
'index': 2,
|
40 |
+
'word': 'تهران',
|
41 |
+
|
42 |
+
sample2:
|
43 |
+
'entity': 'B-pers',
|
44 |
+
'score': 0.99988234,
|
45 |
+
'index': 2,
|
46 |
+
'word': 'عباس',
|