small-yoruba-finetuned-ner
This model is a fine-tuned version of bert-base-multilingual-cased on the wikiann dataset. It achieves the following results on the evaluation set:
- Loss: 0.5450
- Precision: 0.7748
- Recall: 0.7748
- F1: 0.7890
- Accuracy: 0.8967
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: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 15
Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|---|
| No log | 1.0 | 13 | 1.1087 | 0.5043 | 0.5043 | 0.5225 | 0.6713 |
| No log | 2.0 | 26 | 1.0303 | 0.4 | 0.4 | 0.4229 | 0.6297 |
| No log | 3.0 | 39 | 0.7622 | 0.6147 | 0.6147 | 0.6204 | 0.7456 |
| No log | 4.0 | 52 | 0.8148 | 0.5688 | 0.5688 | 0.5741 | 0.7103 |
| No log | 5.0 | 65 | 0.6816 | 0.6053 | 0.6053 | 0.6244 | 0.7834 |
| No log | 6.0 | 78 | 0.7372 | 0.5826 | 0.5826 | 0.6036 | 0.8048 |
| No log | 7.0 | 91 | 0.5917 | 0.7593 | 0.7593 | 0.7628 | 0.8866 |
| No log | 8.0 | 104 | 0.5758 | 0.7155 | 0.7155 | 0.7444 | 0.8829 |
| No log | 9.0 | 117 | 0.5806 | 0.6903 | 0.6903 | 0.7091 | 0.8741 |
| No log | 10.0 | 130 | 0.5254 | 0.7522 | 0.7522 | 0.7727 | 0.9005 |
| No log | 11.0 | 143 | 0.5422 | 0.7636 | 0.7636 | 0.7742 | 0.8942 |
| No log | 12.0 | 156 | 0.5469 | 0.75 | 0.75 | 0.7671 | 0.8879 |
| No log | 13.0 | 169 | 0.5410 | 0.7890 | 0.7890 | 0.7963 | 0.8942 |
| No log | 14.0 | 182 | 0.5435 | 0.7890 | 0.7890 | 0.7963 | 0.8942 |
| No log | 15.0 | 195 | 0.5450 | 0.7748 | 0.7748 | 0.7890 | 0.8967 |
Framework versions
- Transformers 4.28.1
- Pytorch 2.0.0+cu118
- Datasets 2.12.0
- Tokenizers 0.13.3