Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,87 @@
|
|
| 1 |
-
---
|
| 2 |
-
|
| 3 |
-
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: bert-base-cased
|
| 3 |
+
model-index:
|
| 4 |
+
- name: bert-base-cased-ner-rfb
|
| 5 |
+
results: []
|
| 6 |
+
license: apache-2.0
|
| 7 |
+
language:
|
| 8 |
+
- en
|
| 9 |
+
metrics:
|
| 10 |
+
- accuracy
|
| 11 |
+
- f1
|
| 12 |
+
pipeline_tag: token-classification
|
| 13 |
+
---
|
| 14 |
+
|
| 15 |
+
This model is a fine-tuned version of [bert-base-cased](https://huggingface.co/google-bert/bert-base-cased) on a private dataset.
|
| 16 |
+
|
| 17 |
+
It achieves the following results on the evaluation set:
|
| 18 |
+
- eval_loss: 1.2720
|
| 19 |
+
- eval_FILL_precision: 0.7627
|
| 20 |
+
- eval_FILL_recall: 0.7759
|
| 21 |
+
- eval_FILL_f1: 0.7692
|
| 22 |
+
- eval_FILL_number: 58
|
| 23 |
+
- eval_ROLE_precision: 0.8125
|
| 24 |
+
- eval_ROLE_recall: 0.8125
|
| 25 |
+
- eval_ROLE_f1: 0.8125
|
| 26 |
+
- eval_ROLE_number: 48
|
| 27 |
+
- eval_overall_precision: 0.7850
|
| 28 |
+
- eval_overall_recall: 0.7925
|
| 29 |
+
- eval_overall_f1: 0.7887
|
| 30 |
+
- eval_overall_accuracy: 0.8289
|
| 31 |
+
- eval_runtime: 1.3592
|
| 32 |
+
- eval_samples_per_second: 44.144
|
| 33 |
+
- eval_steps_per_second: 5.886
|
| 34 |
+
- step: 0
|
| 35 |
+
|
| 36 |
+
It achieves the following results on the test set:
|
| 37 |
+
- test_FILL_f1: 0.8039
|
| 38 |
+
- test_FILL_number: 46,
|
| 39 |
+
- test_FILL_precision: 0.7321
|
| 40 |
+
- test_FILL_recall: 0.8913
|
| 41 |
+
- test_ROLE_f1: 0.8182
|
| 42 |
+
- test_ROLE_number: 42,
|
| 43 |
+
- test_ROLE_precision: 0.7826
|
| 44 |
+
- test_ROLE_recall: 0.8571
|
| 45 |
+
- test_loss: 0.9132
|
| 46 |
+
- test_overall_accuracy: 0.8791
|
| 47 |
+
- test_overall_f1: 0.8105
|
| 48 |
+
- test_overall_precision: 0.7549
|
| 49 |
+
- test_overall_recall: 0.875
|
| 50 |
+
- test_runtime: 0.9583
|
| 51 |
+
- test_samples_per_second: 63.652
|
| 52 |
+
- test_steps_per_second: 8.348
|
| 53 |
+
|
| 54 |
+
|
| 55 |
+
## Model description
|
| 56 |
+
|
| 57 |
+
More information needed
|
| 58 |
+
|
| 59 |
+
## Intended uses & limitations
|
| 60 |
+
|
| 61 |
+
More information needed
|
| 62 |
+
|
| 63 |
+
## Training and evaluation data
|
| 64 |
+
|
| 65 |
+
More information needed
|
| 66 |
+
|
| 67 |
+
## Training procedure
|
| 68 |
+
|
| 69 |
+
### Training hyperparameters
|
| 70 |
+
|
| 71 |
+
The following hyperparameters were used during training:
|
| 72 |
+
- learning_rate: 0.0001
|
| 73 |
+
- train_batch_size: 8
|
| 74 |
+
- eval_batch_size: 8
|
| 75 |
+
- seed: 42
|
| 76 |
+
- gradient_accumulation_steps: 2
|
| 77 |
+
- total_train_batch_size: 16
|
| 78 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 79 |
+
- lr_scheduler_type: linear
|
| 80 |
+
- training_steps: 600
|
| 81 |
+
|
| 82 |
+
### Framework versions
|
| 83 |
+
|
| 84 |
+
- Transformers 4.40.1
|
| 85 |
+
- Pytorch 2.2.1+cu121
|
| 86 |
+
- Datasets 2.14.6
|
| 87 |
+
- Tokenizers 0.19.1
|