Upload 15 files
Browse files- results/f1_macro.png +0 -0
- results/f1_micro.png +0 -0
- results/f1_weighted.png +0 -0
- results/loss.png +0 -0
- results/precision_macro.png +0 -0
- results/precision_micro.png +0 -0
- results/precision_weighted.png +0 -0
- results/recall_macro.png +0 -0
- results/recall_micro.png +0 -0
- results/recall_weighted.png +0 -0
- results/results_E6.txt +45 -0
- results/results_E7.txt +45 -0
- results/results_E8.txt +45 -0
- results/results_E9.txt +45 -0
- results/syngnn_main.log +352 -0
results/f1_macro.png
ADDED
![]() |
results/f1_micro.png
ADDED
![]() |
results/f1_weighted.png
ADDED
![]() |
results/loss.png
ADDED
![]() |
results/precision_macro.png
ADDED
![]() |
results/precision_micro.png
ADDED
![]() |
results/precision_weighted.png
ADDED
![]() |
results/recall_macro.png
ADDED
![]() |
results/recall_micro.png
ADDED
![]() |
results/recall_weighted.png
ADDED
![]() |
results/results_E6.txt
ADDED
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
***** Test results *****
|
2 |
+
Thu Sep 22 06:41:21 2022
|
3 |
+
Task: ner
|
4 |
+
Model path: bert-base-uncased
|
5 |
+
Data path: ./data/ud/
|
6 |
+
Tokenizer: bert-base-uncased
|
7 |
+
Batch size: 32
|
8 |
+
Epoch: 6
|
9 |
+
Learning rate: 2e-05
|
10 |
+
LR Decay End Factor: 0.3LR Decay End Epoch: 5Sequence length: 96
|
11 |
+
Training: True
|
12 |
+
Num Threads: 24
|
13 |
+
Num Sentences: 0
|
14 |
+
Max Grad Norm: 0.0
|
15 |
+
Use GNN: False
|
16 |
+
Syntax graph style: dep
|
17 |
+
Use label weights: False
|
18 |
+
Clip value: 50
|
19 |
+
precision recall f1-score support
|
20 |
+
|
21 |
+
CARDINAL 0.7133 0.6503 0.6803 612
|
22 |
+
DATE 0.6922 0.7254 0.7084 1045
|
23 |
+
EVENT 0.4429 0.3875 0.4133 80
|
24 |
+
FAC 0.3390 0.3974 0.3659 151
|
25 |
+
GPE 0.8456 0.8714 0.8583 1936
|
26 |
+
LANGUAGE 0.5135 0.2468 0.3333 77
|
27 |
+
LAW 0.4130 0.3333 0.3689 57
|
28 |
+
LOC 0.5934 0.4977 0.5414 217
|
29 |
+
MONEY 0.5370 0.4754 0.5043 61
|
30 |
+
NORP 0.6211 0.7536 0.6809 422
|
31 |
+
ORDINAL 0.8208 0.8304 0.8256 171
|
32 |
+
ORG 0.5289 0.5869 0.5564 857
|
33 |
+
PERCENT 0.3333 0.4722 0.3908 36
|
34 |
+
PERSON 0.7192 0.7885 0.7523 1371
|
35 |
+
PRODUCT 0.2705 0.3367 0.3000 98
|
36 |
+
QUANTITY 0.3485 0.4340 0.3866 53
|
37 |
+
SEP] 0.0000 0.0000 0.0000 0
|
38 |
+
TIME 0.6071 0.6355 0.6210 214
|
39 |
+
WORK_OF_ART 0.3000 0.2538 0.2750 130
|
40 |
+
|
41 |
+
micro avg 0.6487 0.7110 0.6784 7588
|
42 |
+
macro avg 0.5073 0.5093 0.5033 7588
|
43 |
+
weighted avg 0.6821 0.7110 0.6946 7588
|
44 |
+
|
45 |
+
Special token predictions: 0
|
results/results_E7.txt
ADDED
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
***** Test results *****
|
2 |
+
Thu Sep 22 07:44:16 2022
|
3 |
+
Task: ner
|
4 |
+
Model path: bert-base-uncased
|
5 |
+
Data path: ./data/ud/
|
6 |
+
Tokenizer: bert-base-uncased
|
7 |
+
Batch size: 32
|
8 |
+
Epoch: 7
|
9 |
+
Learning rate: 2e-05
|
10 |
+
LR Decay End Factor: 0.3LR Decay End Epoch: 5Sequence length: 96
|
11 |
+
Training: True
|
12 |
+
Num Threads: 24
|
13 |
+
Num Sentences: 0
|
14 |
+
Max Grad Norm: 0.0
|
15 |
+
Use GNN: False
|
16 |
+
Syntax graph style: dep
|
17 |
+
Use label weights: False
|
18 |
+
Clip value: 50
|
19 |
+
precision recall f1-score support
|
20 |
+
|
21 |
+
CARDINAL 0.7030 0.6225 0.6603 612
|
22 |
+
DATE 0.7036 0.7177 0.7106 1045
|
23 |
+
EVENT 0.4133 0.3875 0.4000 80
|
24 |
+
FAC 0.3661 0.4437 0.4012 151
|
25 |
+
GPE 0.8721 0.8667 0.8694 1936
|
26 |
+
LANGUAGE 0.5758 0.2468 0.3455 77
|
27 |
+
LAW 0.3621 0.3684 0.3652 57
|
28 |
+
LOC 0.4978 0.5115 0.5045 217
|
29 |
+
MONEY 0.5849 0.5082 0.5439 61
|
30 |
+
NORP 0.6927 0.7156 0.7040 422
|
31 |
+
ORDINAL 0.8035 0.8129 0.8081 171
|
32 |
+
ORG 0.5158 0.5893 0.5501 857
|
33 |
+
PERCENT 0.3878 0.5278 0.4471 36
|
34 |
+
PERSON 0.7476 0.7994 0.7726 1371
|
35 |
+
PRODUCT 0.2742 0.3469 0.3063 98
|
36 |
+
QUANTITY 0.3443 0.3962 0.3684 53
|
37 |
+
SEP] 0.0000 0.0000 0.0000 0
|
38 |
+
TIME 0.5816 0.6495 0.6137 214
|
39 |
+
WORK_OF_ART 0.3544 0.2154 0.2679 130
|
40 |
+
|
41 |
+
micro avg 0.6680 0.7080 0.6874 7588
|
42 |
+
macro avg 0.5148 0.5119 0.5073 7588
|
43 |
+
weighted avg 0.6955 0.7080 0.6998 7588
|
44 |
+
|
45 |
+
Special token predictions: 0
|
results/results_E8.txt
ADDED
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
***** Test results *****
|
2 |
+
Thu Sep 22 08:47:08 2022
|
3 |
+
Task: ner
|
4 |
+
Model path: bert-base-uncased
|
5 |
+
Data path: ./data/ud/
|
6 |
+
Tokenizer: bert-base-uncased
|
7 |
+
Batch size: 32
|
8 |
+
Epoch: 8
|
9 |
+
Learning rate: 2e-05
|
10 |
+
LR Decay End Factor: 0.3LR Decay End Epoch: 5Sequence length: 96
|
11 |
+
Training: True
|
12 |
+
Num Threads: 24
|
13 |
+
Num Sentences: 0
|
14 |
+
Max Grad Norm: 0.0
|
15 |
+
Use GNN: False
|
16 |
+
Syntax graph style: dep
|
17 |
+
Use label weights: False
|
18 |
+
Clip value: 50
|
19 |
+
precision recall f1-score support
|
20 |
+
|
21 |
+
CARDINAL 0.7269 0.6307 0.6754 612
|
22 |
+
DATE 0.6856 0.7053 0.6953 1045
|
23 |
+
EVENT 0.4286 0.4500 0.4390 80
|
24 |
+
FAC 0.3454 0.4437 0.3884 151
|
25 |
+
GPE 0.8709 0.8574 0.8641 1936
|
26 |
+
LANGUAGE 0.5758 0.2468 0.3455 77
|
27 |
+
LAW 0.4314 0.3860 0.4074 57
|
28 |
+
LOC 0.5829 0.4700 0.5204 217
|
29 |
+
MONEY 0.5085 0.4918 0.5000 61
|
30 |
+
NORP 0.7023 0.7156 0.7089 422
|
31 |
+
ORDINAL 0.8258 0.8596 0.8424 171
|
32 |
+
ORG 0.5300 0.5776 0.5528 857
|
33 |
+
PERCENT 0.4255 0.5556 0.4819 36
|
34 |
+
PERSON 0.7398 0.7841 0.7613 1371
|
35 |
+
PRODUCT 0.2975 0.3673 0.3288 98
|
36 |
+
QUANTITY 0.3284 0.4151 0.3667 53
|
37 |
+
SEP] 0.0000 0.0000 0.0000 0
|
38 |
+
TIME 0.5586 0.6682 0.6085 214
|
39 |
+
WORK_OF_ART 0.3010 0.2385 0.2661 130
|
40 |
+
|
41 |
+
micro avg 0.6607 0.7024 0.6809 7588
|
42 |
+
macro avg 0.5192 0.5191 0.5133 7588
|
43 |
+
weighted avg 0.6968 0.7024 0.6977 7588
|
44 |
+
|
45 |
+
Special token predictions: 0
|
results/results_E9.txt
ADDED
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
***** Test results *****
|
2 |
+
Thu Sep 22 09:50:02 2022
|
3 |
+
Task: ner
|
4 |
+
Model path: bert-base-uncased
|
5 |
+
Data path: ./data/ud/
|
6 |
+
Tokenizer: bert-base-uncased
|
7 |
+
Batch size: 32
|
8 |
+
Epoch: 9
|
9 |
+
Learning rate: 2e-05
|
10 |
+
LR Decay End Factor: 0.3LR Decay End Epoch: 5Sequence length: 96
|
11 |
+
Training: True
|
12 |
+
Num Threads: 24
|
13 |
+
Num Sentences: 0
|
14 |
+
Max Grad Norm: 0.0
|
15 |
+
Use GNN: False
|
16 |
+
Syntax graph style: dep
|
17 |
+
Use label weights: False
|
18 |
+
Clip value: 50
|
19 |
+
precision recall f1-score support
|
20 |
+
|
21 |
+
CARDINAL 0.6809 0.6520 0.6661 612
|
22 |
+
DATE 0.6870 0.7225 0.7043 1045
|
23 |
+
EVENT 0.3977 0.4375 0.4167 80
|
24 |
+
FAC 0.3404 0.4238 0.3776 151
|
25 |
+
GPE 0.8799 0.8549 0.8672 1936
|
26 |
+
LANGUAGE 0.4906 0.3377 0.4000 77
|
27 |
+
LAW 0.4062 0.4561 0.4298 57
|
28 |
+
LOC 0.5000 0.5023 0.5011 217
|
29 |
+
MONEY 0.5161 0.5246 0.5203 61
|
30 |
+
NORP 0.6817 0.7512 0.7148 422
|
31 |
+
ORDINAL 0.8276 0.8421 0.8348 171
|
32 |
+
ORG 0.5455 0.5741 0.5594 857
|
33 |
+
PERCENT 0.5476 0.6389 0.5897 36
|
34 |
+
PERSON 0.7531 0.7943 0.7732 1371
|
35 |
+
PRODUCT 0.2937 0.4286 0.3485 98
|
36 |
+
QUANTITY 0.3492 0.4151 0.3793 53
|
37 |
+
SEP] 0.0000 0.0000 0.0000 0
|
38 |
+
TIME 0.5748 0.6822 0.6239 214
|
39 |
+
WORK_OF_ART 0.2963 0.2462 0.2689 130
|
40 |
+
|
41 |
+
micro avg 0.6736 0.7127 0.6926 7588
|
42 |
+
macro avg 0.5141 0.5413 0.5250 7588
|
43 |
+
weighted avg 0.6958 0.7127 0.7033 7588
|
44 |
+
|
45 |
+
Special token predictions: 0
|
results/syngnn_main.log
ADDED
@@ -0,0 +1,352 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
0 |
0%| | 0/942 [00:00<?, ?it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
0%| | 0/153 [00:00<?, ?it/s]
|
|
|
|
|
|
|
|
|
2 |
0%| | 0/942 [00:00<?, ?it/s]
|
|
|
|
|
|
|
3 |
0%| | 0/153 [00:00<?, ?it/s]
|
|
|
|
|
|
|
|
|
4 |
0%| | 0/942 [00:00<?, ?it/s]
|
|
|
|
|
|
|
5 |
0%| | 0/153 [00:00<?, ?it/s]
|
|
|
|
|
|
|
|
|
6 |
0%| | 0/942 [00:00<?, ?it/s]
|
|
|
|
|
|
|
7 |
0%| | 0/153 [00:00<?, ?it/s]
|
|
|
|
|
|
|
|
|
8 |
0%| | 0/942 [00:00<?, ?it/s]
|
|
|
|
|
|
|
9 |
0%| | 0/153 [00:00<?, ?it/s]
|
|
|
|
|
|
|
|
|
10 |
0%| | 0/942 [00:00<?, ?it/s]
|
|
|
|
|
|
|
11 |
0%| | 0/153 [00:00<?, ?it/s]
|
|
|
|
|
|
|
|
|
12 |
0%| | 0/942 [00:00<?, ?it/s]
|
|
|
|
|
|
|
13 |
0%| | 0/153 [00:00<?, ?it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
14 |
0%| | 0/196 [00:00<?, ?it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
15 |
0%| | 0/942 [00:00<?, ?it/s]
|
|
|
|
|
|
|
16 |
0%| | 0/153 [00:00<?, ?it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
17 |
0%| | 0/196 [00:00<?, ?it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
18 |
0%| | 0/942 [00:00<?, ?it/s]
|
|
|
|
|
|
|
19 |
0%| | 0/153 [00:00<?, ?it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
20 |
0%| | 0/196 [00:00<?, ?it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
21 |
0%| | 0/942 [00:00<?, ?it/s]
|
|
|
|
|
|
|
22 |
0%| | 0/153 [00:00<?, ?it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
23 |
0%| | 0/196 [00:00<?, ?it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
24 |
0%| | 0/196 [00:00<?, ?it/s]
|
|
|
1 |
+
Loading model from path bert-base-uncased
|
2 |
+
Task: ner
|
3 |
+
Model path: bert-base-uncased
|
4 |
+
Data path: ./data/ud/
|
5 |
+
Tokenizer: bert-base-uncased
|
6 |
+
Batch size: 32
|
7 |
+
Epochs: 10
|
8 |
+
Learning rate: 2e-05
|
9 |
+
LR Decay: 0.3
|
10 |
+
LR Decay End Epoch: 5
|
11 |
+
Sequence length: 96
|
12 |
+
Training: True
|
13 |
+
Num Threads: 24
|
14 |
+
Num Sentences: 0
|
15 |
+
Max Norm: 0.0
|
16 |
+
Use GNN: False
|
17 |
+
Use label weights: False
|
18 |
+
PID: 3523179, PGID: 3523174
|
19 |
+
ATen/Parallel:
|
20 |
+
at::get_num_threads() : 24
|
21 |
+
at::get_num_interop_threads() : 36
|
22 |
+
OpenMP 201511 (a.k.a. OpenMP 4.5)
|
23 |
+
omp_get_max_threads() : 24
|
24 |
+
Intel(R) Math Kernel Library Version 2020.0.0 Product Build 20191122 for Intel(R) 64 architecture applications
|
25 |
+
mkl_get_max_threads() : 24
|
26 |
+
Intel(R) MKL-DNN v2.6.0 (Git Hash 52b5f107dd9cf10910aaa19cb47f3abf9b349815)
|
27 |
+
std::thread::hardware_concurrency() : 72
|
28 |
+
Environment variables:
|
29 |
+
OMP_NUM_THREADS : 24
|
30 |
+
MKL_NUM_THREADS : 24
|
31 |
+
ATen parallel backend: OpenMP
|
32 |
+
|
33 |
+
Training model
|
34 |
+
Loading Training Data
|
35 |
+
Loading NER labels from ./data/ud/**/*-train-orig.ner
|
36 |
+
en_atis-ud-train-orig.ner
|
37 |
+
num sentences: 4274
|
38 |
+
en_cesl-ud-train-orig.ner
|
39 |
+
num sentences: 4124
|
40 |
+
en_ewt-ud-train-orig.ner
|
41 |
+
num sentences: 11649
|
42 |
+
en_gum-ud-train-orig.ner
|
43 |
+
num sentences: 5344
|
44 |
+
en_lines-ud-train-orig.ner
|
45 |
+
num sentences: 3010
|
46 |
+
en_partut-ud-train-orig.ner
|
47 |
+
num sentences: 1739
|
48 |
+
Example of NER labels: [[['what', 'O'], ['is', 'O'], ['the', 'O'], ['cost', 'O'], ['of', 'O'], ['a', 'O'], ['round', 'O'], ['trip', 'O'], ['flight', 'O'], ['from', 'O'], ['pittsburgh', 'S-GPE'], ['to', 'O'], ['atlanta', 'S-GPE'], ['beginning', 'O'], ['on', 'O'], ['april', 'B-DATE'], ['twenty', 'I-DATE'], ['fifth', 'E-DATE'], ['and', 'O'], ['returning', 'O'], ['on', 'O'], ['may', 'B-DATE'], ['sixth', 'E-DATE']], [['now', 'O'], ['i', 'O'], ['need', 'O'], ['a', 'O'], ['flight', 'O'], ['leaving', 'O'], ['fort', 'B-GPE'], ['worth', 'E-GPE'], ['and', 'O'], ['arriving', 'O'], ['in', 'O'], ['denver', 'S-GPE'], ['no', 'O'], ['later', 'O'], ['than', 'O'], ['2', 'B-TIME'], ['pm', 'E-TIME'], ['next', 'B-DATE'], ['monday', 'E-DATE']]]
|
49 |
+
30140 sentences, 942 batches of size 32
|
50 |
+
|
51 |
+
Control example of InputFeatures
|
52 |
+
Input Ids: [101, 2085, 1045, 2342, 1037, 3462, 2975, 3481, 4276, 1998, 7194, 1999, 7573, 2053, 2101, 2084, 1016, 7610, 2279, 6928, 102, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
53 |
+
Input Mask: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
54 |
+
Label Ids: [77, 1, 1, 1, 1, 1, 1, 31, 32, 1, 1, 1, 16, 1, 1, 1, 28, 30, 17, 19, 78, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
55 |
+
Valid Ids: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]
|
56 |
+
Label Mask: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
57 |
+
Segment Ids: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
58 |
+
Loading Validation Data
|
59 |
+
Loading NER labels from ./data/ud/**/*-dev-orig.ner
|
60 |
+
en_atis-ud-dev-orig.ner
|
61 |
+
num sentences: 572
|
62 |
+
en_cesl-ud-dev-orig.ner
|
63 |
+
num sentences: 500
|
64 |
+
en_ewt-ud-dev-orig.ner
|
65 |
+
num sentences: 1875
|
66 |
+
en_gum-ud-dev-orig.ner
|
67 |
+
num sentences: 788
|
68 |
+
en_lines-ud-dev-orig.ner
|
69 |
+
num sentences: 986
|
70 |
+
en_partut-ud-dev-orig.ner
|
71 |
+
num sentences: 149
|
72 |
+
Example of NER labels: [[['i', 'O'], ['would', 'O'], ['like', 'O'], ['the', 'O'], ['cheapest', 'O'], ['flight', 'O'], ['from', 'O'], ['pittsburgh', 'S-GPE'], ['to', 'O'], ['atlanta', 'S-GPE'], ['leaving', 'O'], ['april', 'B-DATE'], ['twenty', 'I-DATE'], ['fifth', 'E-DATE'], ['and', 'O'], ['returning', 'O'], ['may', 'B-DATE'], ['sixth', 'E-DATE']], [['i', 'O'], ['want', 'O'], ['a', 'O'], ['flight', 'O'], ['from', 'O'], ['memphis', 'S-LOC'], ['to', 'O'], ['seattle', 'S-FAC'], ['that', 'O'], ['arrives', 'O'], ['no', 'O'], ['later', 'O'], ['than', 'O'], ['3', 'B-TIME'], ['pm', 'E-TIME']]]
|
73 |
+
4870 sentences, 153 batches of size 32
|
74 |
+
|
75 |
+
Control example of InputFeatures
|
76 |
+
Input Ids: [101, 1045, 2215, 1037, 3462, 2013, 9774, 2000, 5862, 2008, 8480, 2053, 2101, 2084, 1017, 7610, 102, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
77 |
+
Input Mask: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
78 |
+
Label Ids: [77, 1, 1, 1, 1, 1, 59, 1, 60, 1, 1, 1, 1, 1, 28, 30, 78, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
79 |
+
Valid Ids: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]
|
80 |
+
Label Mask: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
81 |
+
Segment Ids: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
82 |
+
Test Data
|
83 |
+
Loading NER labels from ./data/ud/**/*-test-orig.ner
|
84 |
+
en_atis-ud-test-orig.ner
|
85 |
+
num sentences: 586
|
86 |
+
en_cesl-ud-test-orig.ner
|
87 |
+
num sentences: 500
|
88 |
+
en_ewt-ud-test-orig.ner
|
89 |
+
num sentences: 1955
|
90 |
+
en_gum-ud-test-orig.ner
|
91 |
+
num sentences: 851
|
92 |
+
en_lines-ud-test-orig.ner
|
93 |
+
num sentences: 988
|
94 |
+
en_pud-ud-test-orig.ner
|
95 |
+
num sentences: 973
|
96 |
+
en_partut-ud-test-orig.ner
|
97 |
+
num sentences: 149
|
98 |
+
en_pronouns-ud-test-orig.ner
|
99 |
+
Some weights of the model checkpoint at bert-base-uncased were not used when initializing BertForNer: ['cls.predictions.transform.dense.weight', 'cls.seq_relationship.bias', 'cls.predictions.transform.LayerNorm.bias', 'cls.seq_relationship.weight', 'cls.predictions.bias', 'cls.predictions.transform.LayerNorm.weight', 'cls.predictions.transform.dense.bias', 'cls.predictions.decoder.weight']
|
100 |
+
- This IS expected if you are initializing BertForNer from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).
|
101 |
+
- This IS NOT expected if you are initializing BertForNer from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).
|
102 |
+
Some weights of BertForNer were not initialized from the model checkpoint at bert-base-uncased and are newly initialized: ['classifier.weight', 'classifier.bias']
|
103 |
+
You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.
|
104 |
+
num sentences: 265
|
105 |
+
Example of NER labels: [[['what', 'O'], ['are', 'O'], ['the', 'O'], ['coach', 'O'], ['flights', 'O'], ['between', 'O'], ['dallas', 'S-GPE'], ['and', 'O'], ['baltimore', 'S-GPE'], ['leaving', 'O'], ['august', 'B-DATE'], ['tenth', 'E-DATE'], ['and', 'O'], ['returning', 'O'], ['august', 'B-DATE'], ['twelve', 'E-DATE']], [['i', 'O'], ['want', 'O'], ['a', 'O'], ['flight', 'O'], ['from', 'O'], ['nashville', 'S-GPE'], ['to', 'O'], ['seattle', 'S-GPE'], ['that', 'O'], ['arrives', 'O'], ['no', 'O'], ['later', 'O'], ['than', 'O'], ['3', 'B-TIME'], ['pm', 'E-TIME']]]
|
106 |
+
6267 sentences, 196 batches of size 32
|
107 |
+
|
108 |
+
Control example of InputFeatures
|
109 |
+
Input Ids: [101, 1045, 2215, 1037, 3462, 2013, 8423, 2000, 5862, 2008, 8480, 2053, 2101, 2084, 1017, 7610, 102, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
110 |
+
Input Mask: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
111 |
+
Label Ids: [77, 1, 1, 1, 1, 1, 16, 1, 16, 1, 1, 1, 1, 1, 28, 30, 78, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
112 |
+
Valid Ids: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]
|
113 |
+
Label Mask: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
114 |
+
Segment Ids: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
115 |
+
Adjusting learning rate of group 0 to 2.0000e-05.
|
116 |
+
|
117 |
0%| | 0/942 [00:00<?, ?it/s]
|
118 |
+
/home/9_QuAnTuM_6/cdaniel/venv_syntrans/lib/python3.8/site-packages/seqeval/metrics/sequence_labeling.py:171: UserWarning: [CLS] seems not to be NE tag.
|
119 |
+
warnings.warn('{} seems not to be NE tag.'.format(chunk))
|
120 |
+
/home/9_QuAnTuM_6/cdaniel/venv_syntrans/lib/python3.8/site-packages/seqeval/metrics/sequence_labeling.py:171: UserWarning: <unk> seems not to be NE tag.
|
121 |
+
warnings.warn('{} seems not to be NE tag.'.format(chunk))
|
122 |
+
/home/9_QuAnTuM_6/cdaniel/venv_syntrans/lib/python3.8/site-packages/seqeval/metrics/sequence_labeling.py:171: UserWarning: <START> seems not to be NE tag.
|
123 |
+
warnings.warn('{} seems not to be NE tag.'.format(chunk))
|
124 |
+
/home/9_QuAnTuM_6/cdaniel/venv_syntrans/lib/python3.8/site-packages/seqeval/metrics/sequence_labeling.py:171: UserWarning: X seems not to be NE tag.
|
125 |
+
warnings.warn('{} seems not to be NE tag.'.format(chunk))
|
126 |
+
/home/9_QuAnTuM_6/cdaniel/venv_syntrans/lib/python3.8/site-packages/seqeval/metrics/sequence_labeling.py:171: UserWarning: [SEP] seems not to be NE tag.
|
127 |
+
warnings.warn('{} seems not to be NE tag.'.format(chunk))
|
128 |
+
/home/9_QuAnTuM_6/cdaniel/venv_syntrans/lib/python3.8/site-packages/seqeval/metrics/sequence_labeling.py:171: UserWarning: <STOP> seems not to be NE tag.
|
129 |
+
warnings.warn('{} seems not to be NE tag.'.format(chunk))
|
130 |
+
O Token Predictions: 471383, NER token predictions: 32921
|
131 |
+
loss: 0.4831624427086608 w prec: 0.48333733331137896 w recall: 0.32263332619404583 w f1: 0.3743943894083346
|
132 |
+
|
133 |
0%| | 0/153 [00:00<?, ?it/s]
|
134 |
+
O Token Predictions: 68594, NER token predictions: 6306
|
135 |
+
loss: 0.24858878246125052 w prec: 0.5736847431314509 w recall: 0.5198167695678152 w f1: 0.5394522935165895
|
136 |
+
Adjusting learning rate of group 0 to 1.7200e-05.
|
137 |
+
|
138 |
0%| | 0/942 [00:00<?, ?it/s]
|
139 |
+
O Token Predictions: 457692, NER token predictions: 46666
|
140 |
+
loss: 0.23743267873828072 w prec: 0.642246843284797 w recall: 0.5536785178839152 w f1: 0.5869654887673703
|
141 |
+
|
142 |
0%| | 0/153 [00:00<?, ?it/s]
|
143 |
+
O Token Predictions: 67004, NER token predictions: 7896
|
144 |
+
loss: 0.19031139838150124 w prec: 0.6106517236837183 w recall: 0.6608245369448317 w f1: 0.6317247629054747
|
145 |
+
Adjusting learning rate of group 0 to 1.4400e-05.
|
146 |
+
|
147 |
0%| | 0/942 [00:00<?, ?it/s]
|
148 |
+
O Token Predictions: 452542, NER token predictions: 51816
|
149 |
+
loss: 0.17240141229723796 w prec: 0.6995943496786741 w recall: 0.6429374598415079 w f1: 0.6678698690045989
|
150 |
+
|
151 |
0%| | 0/153 [00:00<?, ?it/s]
|
152 |
+
O Token Predictions: 67555, NER token predictions: 7345
|
153 |
+
loss: 0.16576223912971472 w prec: 0.6909975902827813 w recall: 0.6671977693686517 w f1: 0.6759967434078691
|
154 |
+
Adjusting learning rate of group 0 to 1.1600e-05.
|
155 |
+
|
156 |
0%| | 0/942 [00:00<?, ?it/s]
|
157 |
+
O Token Predictions: 450026, NER token predictions: 54332
|
158 |
+
loss: 0.13777431711601984 w prec: 0.7394564960771575 w recall: 0.7012208181623474 w f1: 0.718659603222537
|
159 |
+
|
160 |
0%| | 0/153 [00:00<?, ?it/s]
|
161 |
+
O Token Predictions: 67201, NER token predictions: 7699
|
162 |
+
loss: 0.16132921111934326 w prec: 0.6829530578688375 w recall: 0.7128062139016133 w f1: 0.6956742392256572
|
163 |
+
Adjusting learning rate of group 0 to 8.8000e-06.
|
164 |
+
|
165 |
0%| | 0/942 [00:00<?, ?it/s]
|
166 |
+
O Token Predictions: 448629, NER token predictions: 55729
|
167 |
+
loss: 0.11438079141950405 w prec: 0.770823616445863 w recall: 0.7418344399228957 w f1: 0.7551456680915405
|
168 |
+
|
169 |
0%| | 0/153 [00:00<?, ?it/s]
|
170 |
+
O Token Predictions: 67145, NER token predictions: 7755
|
171 |
+
loss: 0.15573614080941756 w prec: 0.720899538634842 w recall: 0.7211710814578769 w f1: 0.7191736846769018
|
172 |
+
Adjusting learning rate of group 0 to 6.0000e-06.
|
173 |
+
|
174 |
0%| | 0/942 [00:00<?, ?it/s]
|
175 |
+
O Token Predictions: 447725, NER token predictions: 56633
|
176 |
+
loss: 0.09880779274579161 w prec: 0.7914035913278309 w recall: 0.7704005140286999 w f1: 0.780219783818005
|
177 |
+
|
178 |
0%| | 0/153 [00:00<?, ?it/s]
|
179 |
+
O Token Predictions: 66994, NER token predictions: 7906
|
180 |
+
loss: 0.15097037561578688 w prec: 0.7114692952948101 w recall: 0.7382991435968931 w f1: 0.7234172615803383
|
181 |
+
Adjusting learning rate of group 0 to 6.0000e-06.
|
182 |
+
|
183 |
0%| | 0/942 [00:00<?, ?it/s]
|
184 |
+
O Token Predictions: 447405, NER token predictions: 56953
|
185 |
+
loss: 0.0908111283029657 w prec: 0.8023543679818423 w recall: 0.7863300492610837 w f1: 0.7937841209593911
|
186 |
+
|
187 |
0%| | 0/153 [00:00<?, ?it/s]
|
188 |
+
loss: 0.15317525608112026 w prec: 0.7115102285753689 w recall: 0.7444732125074687 w f1: 0.7263377157668733
|
189 |
+
Adjusting learning rate of group 0 to 6.0000e-06.
|
190 |
+
Model evaluation
|
191 |
+
|
192 |
+
|
193 |
+
|
194 |
0%| | 0/196 [00:00<?, ?it/s]
|
195 |
+
/home/9_QuAnTuM_6/cdaniel/venv_syntrans/lib/python3.8/site-packages/seqeval/metrics/v1.py:57: UndefinedMetricWarning: Recall and F-score are ill-defined and being set to 0.0 in labels with no true samples. Use `zero_division` parameter to control this behavior.
|
196 |
+
_warn_prf(average, modifier, msg_start, len(result))
|
197 |
+
***** Test results *****
|
198 |
+
precision recall f1-score support
|
199 |
+
|
200 |
+
CARDINAL 0.7133 0.6503 0.6803 612
|
201 |
+
DATE 0.6922 0.7254 0.7084 1045
|
202 |
+
EVENT 0.4429 0.3875 0.4133 80
|
203 |
+
FAC 0.3390 0.3974 0.3659 151
|
204 |
+
GPE 0.8456 0.8714 0.8583 1936
|
205 |
+
LANGUAGE 0.5135 0.2468 0.3333 77
|
206 |
+
LAW 0.4130 0.3333 0.3689 57
|
207 |
+
LOC 0.5934 0.4977 0.5414 217
|
208 |
+
MONEY 0.5370 0.4754 0.5043 61
|
209 |
+
NORP 0.6211 0.7536 0.6809 422
|
210 |
+
ORDINAL 0.8208 0.8304 0.8256 171
|
211 |
+
ORG 0.5289 0.5869 0.5564 857
|
212 |
+
PERCENT 0.3333 0.4722 0.3908 36
|
213 |
+
PERSON 0.7192 0.7885 0.7523 1371
|
214 |
+
PRODUCT 0.2705 0.3367 0.3000 98
|
215 |
+
QUANTITY 0.3485 0.4340 0.3866 53
|
216 |
+
SEP] 0.0000 0.0000 0.0000 0
|
217 |
+
TIME 0.6071 0.6355 0.6210 214
|
218 |
+
WORK_OF_ART 0.3000 0.2538 0.2750 130
|
219 |
+
|
220 |
+
micro avg 0.6487 0.7110 0.6784 7588
|
221 |
+
macro avg 0.5073 0.5093 0.5033 7588
|
222 |
+
weighted avg 0.6821 0.7110 0.6946 7588
|
223 |
+
|
224 |
+
Special token predictions: 0
|
225 |
+
|
226 |
0%| | 0/942 [00:00<?, ?it/s]
|
227 |
+
O Token Predictions: 447026, NER token predictions: 57332
|
228 |
+
loss: 0.0834173557759834 w prec: 0.815729338081138 w recall: 0.8016705932747912 w f1: 0.8082084594984448
|
229 |
+
|
230 |
0%| | 0/153 [00:00<?, ?it/s]
|
231 |
+
loss: 0.15407624415249802 w prec: 0.7284351035722969 w recall: 0.7444732125074687 w f1: 0.7346851885936531
|
232 |
+
Adjusting learning rate of group 0 to 6.0000e-06.
|
233 |
+
Model evaluation
|
234 |
+
|
235 |
+
|
236 |
+
|
237 |
0%| | 0/196 [00:00<?, ?it/s]
|
238 |
+
***** Test results *****
|
239 |
+
precision recall f1-score support
|
240 |
+
|
241 |
+
CARDINAL 0.7030 0.6225 0.6603 612
|
242 |
+
DATE 0.7036 0.7177 0.7106 1045
|
243 |
+
EVENT 0.4133 0.3875 0.4000 80
|
244 |
+
FAC 0.3661 0.4437 0.4012 151
|
245 |
+
GPE 0.8721 0.8667 0.8694 1936
|
246 |
+
LANGUAGE 0.5758 0.2468 0.3455 77
|
247 |
+
LAW 0.3621 0.3684 0.3652 57
|
248 |
+
LOC 0.4978 0.5115 0.5045 217
|
249 |
+
MONEY 0.5849 0.5082 0.5439 61
|
250 |
+
NORP 0.6927 0.7156 0.7040 422
|
251 |
+
ORDINAL 0.8035 0.8129 0.8081 171
|
252 |
+
ORG 0.5158 0.5893 0.5501 857
|
253 |
+
PERCENT 0.3878 0.5278 0.4471 36
|
254 |
+
PERSON 0.7476 0.7994 0.7726 1371
|
255 |
+
PRODUCT 0.2742 0.3469 0.3063 98
|
256 |
+
QUANTITY 0.3443 0.3962 0.3684 53
|
257 |
+
SEP] 0.0000 0.0000 0.0000 0
|
258 |
+
TIME 0.5816 0.6495 0.6137 214
|
259 |
+
WORK_OF_ART 0.3544 0.2154 0.2679 130
|
260 |
+
|
261 |
+
micro avg 0.6680 0.7080 0.6874 7588
|
262 |
+
macro avg 0.5148 0.5119 0.5073 7588
|
263 |
+
weighted avg 0.6955 0.7080 0.6998 7588
|
264 |
+
|
265 |
+
Special token predictions: 0
|
266 |
+
|
267 |
0%| | 0/942 [00:00<?, ?it/s]
|
268 |
+
O Token Predictions: 446823, NER token predictions: 57535
|
269 |
+
loss: 0.07700805802633807 w prec: 0.8281029278359268 w recall: 0.8139590918826302 w f1: 0.8206010130635916
|
270 |
+
|
271 |
0%| | 0/153 [00:00<?, ?it/s]
|
272 |
+
loss: 0.1566143021888398 w prec: 0.7311165058086735 w recall: 0.7394941246763593 w f1: 0.7335244139037927
|
273 |
+
Adjusting learning rate of group 0 to 6.0000e-06.
|
274 |
+
Model evaluation
|
275 |
+
|
276 |
+
|
277 |
+
|
278 |
0%| | 0/196 [00:00<?, ?it/s]
|
279 |
+
***** Test results *****
|
280 |
+
precision recall f1-score support
|
281 |
+
|
282 |
+
CARDINAL 0.7269 0.6307 0.6754 612
|
283 |
+
DATE 0.6856 0.7053 0.6953 1045
|
284 |
+
EVENT 0.4286 0.4500 0.4390 80
|
285 |
+
FAC 0.3454 0.4437 0.3884 151
|
286 |
+
GPE 0.8709 0.8574 0.8641 1936
|
287 |
+
LANGUAGE 0.5758 0.2468 0.3455 77
|
288 |
+
LAW 0.4314 0.3860 0.4074 57
|
289 |
+
LOC 0.5829 0.4700 0.5204 217
|
290 |
+
MONEY 0.5085 0.4918 0.5000 61
|
291 |
+
NORP 0.7023 0.7156 0.7089 422
|
292 |
+
ORDINAL 0.8258 0.8596 0.8424 171
|
293 |
+
ORG 0.5300 0.5776 0.5528 857
|
294 |
+
PERCENT 0.4255 0.5556 0.4819 36
|
295 |
+
PERSON 0.7398 0.7841 0.7613 1371
|
296 |
+
PRODUCT 0.2975 0.3673 0.3288 98
|
297 |
+
QUANTITY 0.3284 0.4151 0.3667 53
|
298 |
+
SEP] 0.0000 0.0000 0.0000 0
|
299 |
+
TIME 0.5586 0.6682 0.6085 214
|
300 |
+
WORK_OF_ART 0.3010 0.2385 0.2661 130
|
301 |
+
|
302 |
+
micro avg 0.6607 0.7024 0.6809 7588
|
303 |
+
macro avg 0.5192 0.5191 0.5133 7588
|
304 |
+
weighted avg 0.6968 0.7024 0.6977 7588
|
305 |
+
|
306 |
+
Special token predictions: 0
|
307 |
+
|
308 |
0%| | 0/942 [00:00<?, ?it/s]
|
309 |
+
O Token Predictions: 446643, NER token predictions: 57715
|
310 |
+
loss: 0.07239252862554131 w prec: 0.8371405381680805 w recall: 0.8260066395373742 w f1: 0.8311608710414287
|
311 |
+
|
312 |
0%| | 0/153 [00:00<?, ?it/s]
|
313 |
+
loss: 0.15688133854540734 w prec: 0.7230917519740736 w recall: 0.7476598287193786 w f1: 0.7338068025133403
|
314 |
+
Adjusting learning rate of group 0 to 6.0000e-06.
|
315 |
+
Model evaluation
|
316 |
+
|
317 |
+
|
318 |
+
|
319 |
0%| | 0/196 [00:00<?, ?it/s]
|
320 |
+
precision recall f1-score support
|
321 |
+
|
322 |
+
CARDINAL 0.6809 0.6520 0.6661 612
|
323 |
+
DATE 0.6870 0.7225 0.7043 1045
|
324 |
+
EVENT 0.3977 0.4375 0.4167 80
|
325 |
+
FAC 0.3404 0.4238 0.3776 151
|
326 |
+
GPE 0.8799 0.8549 0.8672 1936
|
327 |
+
LANGUAGE 0.4906 0.3377 0.4000 77
|
328 |
+
LAW 0.4062 0.4561 0.4298 57
|
329 |
+
LOC 0.5000 0.5023 0.5011 217
|
330 |
+
MONEY 0.5161 0.5246 0.5203 61
|
331 |
+
NORP 0.6817 0.7512 0.7148 422
|
332 |
+
ORDINAL 0.8276 0.8421 0.8348 171
|
333 |
+
ORG 0.5455 0.5741 0.5594 857
|
334 |
+
PERCENT 0.5476 0.6389 0.5897 36
|
335 |
+
PERSON 0.7531 0.7943 0.7732 1371
|
336 |
+
PRODUCT 0.2937 0.4286 0.3485 98
|
337 |
+
QUANTITY 0.3492 0.4151 0.3793 53
|
338 |
+
SEP] 0.0000 0.0000 0.0000 0
|
339 |
+
TIME 0.5748 0.6822 0.6239 214
|
340 |
+
WORK_OF_ART 0.2963 0.2462 0.2689 130
|
341 |
+
|
342 |
+
micro avg 0.6736 0.7127 0.6926 7588
|
343 |
+
macro avg 0.5141 0.5413 0.5250 7588
|
344 |
+
weighted avg 0.6958 0.7127 0.7033 7588
|
345 |
+
|
346 |
+
Special token predictions: 0
|
347 |
+
Test Data
|
348 |
+
Loading NER labels from ./data/ud/**/*-test-orig.ner
|
349 |
+
en_atis-ud-test-orig.ner
|
350 |
+
num sentences: 586
|
351 |
+
en_cesl-ud-test-orig.ner
|
352 |
+
num sentences: 500
|
353 |
+
en_ewt-ud-test-orig.ner
|
354 |
+
num sentences: 1955
|
355 |
+
en_gum-ud-test-orig.ner
|
356 |
+
num sentences: 851
|
357 |
+
en_lines-ud-test-orig.ner
|
358 |
+
num sentences: 988
|
359 |
+
en_pud-ud-test-orig.ner
|
360 |
+
num sentences: 973
|
361 |
+
en_partut-ud-test-orig.ner
|
362 |
+
num sentences: 149
|
363 |
+
en_pronouns-ud-test-orig.ner
|
364 |
+
|
365 |
+
num sentences: 265
|
366 |
+
Example of NER labels: [[['what', 'O'], ['are', 'O'], ['the', 'O'], ['coach', 'O'], ['flights', 'O'], ['between', 'O'], ['dallas', 'S-GPE'], ['and', 'O'], ['baltimore', 'S-GPE'], ['leaving', 'O'], ['august', 'B-DATE'], ['tenth', 'E-DATE'], ['and', 'O'], ['returning', 'O'], ['august', 'B-DATE'], ['twelve', 'E-DATE']], [['i', 'O'], ['want', 'O'], ['a', 'O'], ['flight', 'O'], ['from', 'O'], ['nashville', 'S-GPE'], ['to', 'O'], ['seattle', 'S-GPE'], ['that', 'O'], ['arrives', 'O'], ['no', 'O'], ['later', 'O'], ['than', 'O'], ['3', 'B-TIME'], ['pm', 'E-TIME']]]
|
367 |
+
6267 sentences, 196 batches of size 32
|
368 |
+
|
369 |
+
Control example of InputFeatures
|
370 |
+
Input Ids: [101, 1045, 2215, 1037, 3462, 2013, 8423, 2000, 5862, 2008, 8480, 2053, 2101, 2084, 1017, 7610, 102, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
371 |
+
Input Mask: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
372 |
+
Label Ids: [77, 1, 1, 1, 1, 1, 16, 1, 16, 1, 1, 1, 1, 1, 28, 30, 78, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
373 |
+
Valid Ids: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]
|
374 |
+
Label Mask: [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
375 |
+
Segment Ids: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
|
376 |
+
|
377 |
0%| | 0/196 [00:00<?, ?it/s]
|