Commit
•
25dc135
1
Parent(s):
ccdf3e0
Update README.md
Browse files
README.md
CHANGED
|
@@ -61,14 +61,22 @@ digitized documents from Finnish public administration was also used for model t
|
|
| 61 |
entity classes contained in training, validation and test datasets are listed below:
|
| 62 |
|
| 63 |
Number of entity types in the data
|
| 64 |
-
Dataset|
|
| 65 |
-
|
| 66 |
-
Train|0|0|0|0|0|0|0|0|0|0
|
| 67 |
-
Val|
|
| 68 |
-
Test|
|
| 69 |
|
| 70 |
## Training procedure
|
| 71 |
|
| 72 |
This model was trained using a NVIDIA RTX A6000 GPU with the following hyperparameters:
|
| 73 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 74 |
The training code with instructions is available [here](https://github.com/DALAI-hanke/BERT_NER).
|
|
|
|
| 61 |
entity classes contained in training, validation and test datasets are listed below:
|
| 62 |
|
| 63 |
Number of entity types in the data
|
| 64 |
+
Dataset|PERSON|ORG|LOC|GPE|PRODUCT|EVENT|DATE|JON|FIBC|NORP
|
| 65 |
+
-|-|-|-|-|-|-|-|-|-|-
|
| 66 |
+
Train|0|0|0|0|0|0|0|0|0|0
|
| 67 |
+
Val|1560|4077|108|1643|880|165|1897|185|265|299
|
| 68 |
+
Test|1284|3742|87|1713|906|137|1864|179|234|261
|
| 69 |
|
| 70 |
## Training procedure
|
| 71 |
|
| 72 |
This model was trained using a NVIDIA RTX A6000 GPU with the following hyperparameters:
|
| 73 |
|
| 74 |
+
- learning rate: 2e-05
|
| 75 |
+
- train batch size: 16
|
| 76 |
+
- epochs: 10
|
| 77 |
+
- optimizer: AdamW with betas=(0.9,0.999) and epsilon=1e-08
|
| 78 |
+
- scheduler: linear scheduler with num_warmup_steps=round(len(train_dataloader)/5) and num_training_steps=len(train_dataloader)*epochs
|
| 79 |
+
- maximum length of data sequence: 512
|
| 80 |
+
- patience: 2 epochs
|
| 81 |
+
|
| 82 |
The training code with instructions is available [here](https://github.com/DALAI-hanke/BERT_NER).
|