M. Megahed commited on
Commit
fd75eb2
2 Parent(s): b38c825 6cdd2d5

Merge branch 'main' of https://huggingface.co/megantosh/flair-arabic-multi-ner into main

Browse files
Files changed (1) hide show
  1. README.md +55 -0
README.md ADDED
@@ -0,0 +1,55 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language: ar
3
+ license: apache-2.0
4
+ datasets:
5
+ - AQMAR
6
+ ---
7
+ # Arabic NER Model using Flair Embeddings
8
+ Training was conducted over 94 epochs, using a linear decaying learning rate of 2e-05, and a total batch size of 32.
9
+ 11
10
+ 12
11
+
12
+ Results:
13
+ - F1-score (micro) 0.8666
14
+ - F1-score (macro) 0.8488
15
+
16
+ By class:
17
+ LOC tp: 539 - fp: 51 - fn: 68 - precision: 0.9136 - recall: 0.8880 - f1-score: 0.9006
18
+ MISC tp: 408 - fp: 57 - fn: 89 - precision: 0.8774 - recall: 0.8209 - f1-score: 0.8482
19
+ ORG tp: 167 - fp: 43 - fn: 64 - precision: 0.7952 - recall: 0.7229 - f1-score: 0.7574
20
+ PER tp: 501 - fp: 65 - fn: 60 - precision: 0.8852 - recall: 0.8930 - f1-score: 0.8891
21
+
22
+
23
+ ---
24
+
25
+ ```
26
+ 2020-10-27 12:05:47,801 Model: "SequenceTagger(
27
+ (embeddings): StackedEmbeddings(
28
+ (list_embedding_0): WordEmbeddings('glove')
29
+ (list_embedding_1): FlairEmbeddings(
30
+ (lm): LanguageModel(
31
+ (drop): Dropout(p=0.1, inplace=False)
32
+ (encoder): Embedding(7125, 100)
33
+ (rnn): LSTM(100, 2048)
34
+ (decoder): Linear(in_features=2048, out_features=7125, bias=True)
35
+ )
36
+ )
37
+ (list_embedding_2): FlairEmbeddings(
38
+ (lm): LanguageModel(
39
+ (drop): Dropout(p=0.1, inplace=False)
40
+ (encoder): Embedding(7125, 100)
41
+ (rnn): LSTM(100, 2048)
42
+ (decoder): Linear(in_features=2048, out_features=7125, bias=True)
43
+ )
44
+ )
45
+ )
46
+ (word_dropout): WordDropout(p=0.05)
47
+ (locked_dropout): LockedDropout(p=0.5)
48
+ (embedding2nn): Linear(in_features=4196, out_features=4196, bias=True)
49
+ (rnn): LSTM(4196, 256, batch_first=True, bidirectional=True)
50
+ (linear): Linear(in_features=512, out_features=15, bias=True)
51
+ (beta): 1.0
52
+ (weights): None
53
+ (weight_tensor) None
54
+
55
+ ```