Hugging Face's logo

denizspynk
/
req_mod_ner_modelv2

Token Classification
Transformers
PyTorch
Dutch
roberta
Generated from Trainer
Model card Files Community
1
denizspynk commited on
Commit
9075b99
·
1 Parent(s): 07c1ac8

update model card README.md

Browse files
Files changed (1) hide show
  1. README.md +24 -40
README.md CHANGED
@@ -19,11 +19,11 @@ should probably proofread and complete it, then remove this comment. -->
19
 
20
  This model is a fine-tuned version of [pdelobelle/robbert-v2-dutch-ner](https://huggingface.co/pdelobelle/robbert-v2-dutch-ner) on the None dataset.
21
  It achieves the following results on the evaluation set:
22
- - Loss: 0.7795
23
- - Precision: 0.5932
24
- - Recall: 0.6034
25
- - F1: 0.5983
26
- - Accuracy: 0.9193
27
 
28
  ## Model description
29
 
@@ -43,49 +43,33 @@ More information needed
43
 
44
  The following hyperparameters were used during training:
45
  - learning_rate: 0.0001
46
- - train_batch_size: 2
47
- - eval_batch_size: 2
48
  - seed: 42
49
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
50
  - lr_scheduler_type: linear
51
- - num_epochs: 32
52
 
53
  ### Training results
54
 
55
  | Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
56
  |:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|:------:|:--------:|
57
- | No log | 1.0 | 120 | 0.5525 | 0.6538 | 0.2931 | 0.4048 | 0.9007 |
58
- | No log | 2.0 | 240 | 0.3596 | 0.5447 | 0.5776 | 0.5607 | 0.9120 |
59
- | No log | 3.0 | 360 | 0.4229 | 0.5495 | 0.4310 | 0.4831 | 0.9031 |
60
- | No log | 4.0 | 480 | 0.4091 | 0.5435 | 0.6466 | 0.5906 | 0.9185 |
61
- | 0.2667 | 5.0 | 600 | 0.5476 | 0.6837 | 0.5776 | 0.6262 | 0.9233 |
62
- | 0.2667 | 6.0 | 720 | 0.4703 | 0.5610 | 0.5948 | 0.5774 | 0.9185 |
63
- | 0.2667 | 7.0 | 840 | 0.5904 | 0.5897 | 0.5948 | 0.5923 | 0.9185 |
64
- | 0.2667 | 8.0 | 960 | 0.6285 | 0.5772 | 0.6121 | 0.5941 | 0.9177 |
65
- | 0.0436 | 9.0 | 1080 | 0.7077 | 0.6095 | 0.5517 | 0.5792 | 0.9153 |
66
- | 0.0436 | 10.0 | 1200 | 0.6974 | 0.5929 | 0.5776 | 0.5852 | 0.9177 |
67
- | 0.0436 | 11.0 | 1320 | 0.6777 | 0.5205 | 0.6552 | 0.5802 | 0.9104 |
68
- | 0.0436 | 12.0 | 1440 | 0.6601 | 0.6174 | 0.6121 | 0.6147 | 0.9201 |
69
- | 0.0216 | 13.0 | 1560 | 0.6536 | 0.5809 | 0.6810 | 0.6270 | 0.9209 |
70
- | 0.0216 | 14.0 | 1680 | 0.7329 | 0.5571 | 0.6724 | 0.6094 | 0.9153 |
71
- | 0.0216 | 15.0 | 1800 | 0.7276 | 0.6809 | 0.5517 | 0.6095 | 0.9201 |
72
- | 0.0216 | 16.0 | 1920 | 0.7243 | 0.6017 | 0.6121 | 0.6068 | 0.9209 |
73
- | 0.0164 | 17.0 | 2040 | 0.6963 | 0.592 | 0.6379 | 0.6141 | 0.9217 |
74
- | 0.0164 | 18.0 | 2160 | 0.7418 | 0.6071 | 0.5862 | 0.5965 | 0.9209 |
75
- | 0.0164 | 19.0 | 2280 | 0.8015 | 0.6667 | 0.5690 | 0.6140 | 0.9241 |
76
- | 0.0164 | 20.0 | 2400 | 0.7075 | 0.5168 | 0.6638 | 0.5811 | 0.9136 |
77
- | 0.0098 | 21.0 | 2520 | 0.7847 | 0.6262 | 0.5776 | 0.6009 | 0.9201 |
78
- | 0.0098 | 22.0 | 2640 | 0.7588 | 0.5812 | 0.5862 | 0.5837 | 0.9177 |
79
- | 0.0098 | 23.0 | 2760 | 0.7439 | 0.5530 | 0.6293 | 0.5887 | 0.9153 |
80
- | 0.0098 | 24.0 | 2880 | 0.7619 | 0.5932 | 0.6034 | 0.5983 | 0.9169 |
81
- | 0.0067 | 25.0 | 3000 | 0.7605 | 0.5948 | 0.5948 | 0.5948 | 0.9201 |
82
- | 0.0067 | 26.0 | 3120 | 0.7635 | 0.5847 | 0.5948 | 0.5897 | 0.9185 |
83
- | 0.0067 | 27.0 | 3240 | 0.7732 | 0.6106 | 0.5948 | 0.6026 | 0.9193 |
84
- | 0.0067 | 28.0 | 3360 | 0.7727 | 0.5897 | 0.5948 | 0.5923 | 0.9185 |
85
- | 0.0067 | 29.0 | 3480 | 0.7748 | 0.6 | 0.5948 | 0.5974 | 0.9201 |
86
- | 0.006 | 30.0 | 3600 | 0.7759 | 0.5932 | 0.6034 | 0.5983 | 0.9193 |
87
- | 0.006 | 31.0 | 3720 | 0.7792 | 0.5932 | 0.6034 | 0.5983 | 0.9193 |
88
- | 0.006 | 32.0 | 3840 | 0.7795 | 0.5932 | 0.6034 | 0.5983 | 0.9193 |
89
 
90
 
91
  ### Framework versions
 
19
 
20
  This model is a fine-tuned version of [pdelobelle/robbert-v2-dutch-ner](https://huggingface.co/pdelobelle/robbert-v2-dutch-ner) on the None dataset.
21
  It achieves the following results on the evaluation set:
22
+ - Loss: 0.7115
23
+ - Precision: 0.6522
24
+ - Recall: 0.6466
25
+ - F1: 0.6494
26
+ - Accuracy: 0.9249
27
 
28
  ## Model description
29
 
 
43
 
44
  The following hyperparameters were used during training:
45
  - learning_rate: 0.0001
46
+ - train_batch_size: 1
47
+ - eval_batch_size: 1
48
  - seed: 42
49
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
50
  - lr_scheduler_type: linear
51
+ - num_epochs: 16
52
 
53
  ### Training results
54
 
55
  | Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
56
  |:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|:------:|:--------:|
57
+ | No log | 1.0 | 240 | 0.4780 | 0.3456 | 0.4052 | 0.3730 | 0.8789 |
58
+ | No log | 2.0 | 480 | 0.3903 | 0.5934 | 0.4655 | 0.5217 | 0.9080 |
59
+ | 0.4168 | 3.0 | 720 | 0.5082 | 0.6782 | 0.5086 | 0.5813 | 0.9169 |
60
+ | 0.4168 | 4.0 | 960 | 0.4307 | 0.5846 | 0.6552 | 0.6179 | 0.9201 |
61
+ | 0.1633 | 5.0 | 1200 | 0.5179 | 0.6 | 0.5948 | 0.5974 | 0.9233 |
62
+ | 0.1633 | 6.0 | 1440 | 0.6073 | 0.5752 | 0.5603 | 0.5677 | 0.9185 |
63
+ | 0.0676 | 7.0 | 1680 | 0.6198 | 0.6638 | 0.6638 | 0.6638 | 0.9233 |
64
+ | 0.0676 | 8.0 | 1920 | 0.6876 | 0.6311 | 0.6638 | 0.6471 | 0.9185 |
65
+ | 0.0445 | 9.0 | 2160 | 0.7112 | 0.6522 | 0.6466 | 0.6494 | 0.9201 |
66
+ | 0.0445 | 10.0 | 2400 | 0.7232 | 0.6522 | 0.6466 | 0.6494 | 0.9193 |
67
+ | 0.0259 | 11.0 | 2640 | 0.6511 | 0.6371 | 0.6810 | 0.6583 | 0.9233 |
68
+ | 0.0259 | 12.0 | 2880 | 0.6733 | 0.6783 | 0.6724 | 0.6753 | 0.9257 |
69
+ | 0.0146 | 13.0 | 3120 | 0.6636 | 0.6695 | 0.6810 | 0.6752 | 0.9282 |
70
+ | 0.0146 | 14.0 | 3360 | 0.6943 | 0.6496 | 0.6552 | 0.6524 | 0.9257 |
71
+ | 0.0134 | 15.0 | 3600 | 0.7055 | 0.6552 | 0.6552 | 0.6552 | 0.9257 |
72
+ | 0.0134 | 16.0 | 3840 | 0.7115 | 0.6522 | 0.6466 | 0.6494 | 0.9249 |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
73
 
74
 
75
  ### Framework versions