bert-large-uncased_ner_wnut_17
This model is a fine-tuned version of bert-large-uncased on the wnut_17 dataset. It achieves the following results on the evaluation set:
- Loss: 0.2516
- Precision: 0.7053
- Recall: 0.5754
- F1: 0.6337
- Accuracy: 0.9603
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- num_epochs: 5
Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|---|
| No log | 1.0 | 213 | 0.2143 | 0.6353 | 0.4605 | 0.5340 | 0.9490 |
| No log | 2.0 | 426 | 0.2299 | 0.7322 | 0.5036 | 0.5967 | 0.9556 |
| 0.1489 | 3.0 | 639 | 0.2137 | 0.6583 | 0.5945 | 0.6248 | 0.9603 |
| 0.1489 | 4.0 | 852 | 0.2494 | 0.7035 | 0.5789 | 0.6352 | 0.9604 |
| 0.0268 | 5.0 | 1065 | 0.2516 | 0.7053 | 0.5754 | 0.6337 | 0.9603 |
Framework versions
- Transformers 4.20.1
- Pytorch 1.11.0
- Datasets 2.1.0
- Tokenizers 0.12.1