Update README.md
Browse files
README.md
CHANGED
@@ -34,14 +34,32 @@ This model is a fine-tuned version of [bert-base-multilingual-uncased](https://h
|
|
34 |
| Recall | **94.18** |
|
35 |
|
36 |
|
37 |
-
|
38 |
|
39 |
-
|
40 |
-
-
|
41 |
-
-
|
|
|
42 |
|
43 |
|
44 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
45 |
|
46 |
- Transformers 4.18.0
|
47 |
- TensorFlow 2.8.0
|
@@ -50,3 +68,4 @@ The following hyperparameters were used during training:
|
|
50 |
|
51 |
- [Fine-tune on NER script provided by @stefan-it](https://raw.githubusercontent.com/stefan-it/fine-tuned-berts-seq/master/scripts/preprocess.py)
|
52 |
|
|
|
|
34 |
| Recall | **94.18** |
|
35 |
|
36 |
|
37 |
+
## Training hyperparameters
|
38 |
|
39 |
+
- Optimizer = AdamW
|
40 |
+
- LearningRate = 4e-5
|
41 |
+
- WeightDecay = 1e-2
|
42 |
+
- Warmup = 500
|
43 |
|
44 |
|
45 |
+
## Example of usage
|
46 |
+
|
47 |
+
```python
|
48 |
+
# pipeline example
|
49 |
+
|
50 |
+
from transformers import pipeline
|
51 |
+
|
52 |
+
model_checkpoint = "Sultannn/bert-base-ft-ner-xtreme-id"
|
53 |
+
token_classifier = pipeline(
|
54 |
+
"token-classification", model=model_checkpoint, aggregation_strategy="simple")
|
55 |
+
|
56 |
+
text = "nama saya Tono saya bekerja di Facebook dan tinggal di Jawa"
|
57 |
+
|
58 |
+
token_classifier(text)
|
59 |
+
|
60 |
+
```
|
61 |
+
|
62 |
+
## Framework versions
|
63 |
|
64 |
- Transformers 4.18.0
|
65 |
- TensorFlow 2.8.0
|
|
|
68 |
|
69 |
- [Fine-tune on NER script provided by @stefan-it](https://raw.githubusercontent.com/stefan-it/fine-tuned-berts-seq/master/scripts/preprocess.py)
|
70 |
|
71 |
+
> Made with <span style="color: #e25555;">♥</span> in Paris
|