multilingual-Davlan/afro-xlmr-base-hausa-ner-v1
This model is a fine-tuned version of Davlan/afro-xlmr-base on the Beijuka/Multilingual_PII_NER_dataset dataset. It achieves the following results on the evaluation set:
- Loss: 0.1152
- Precision: 0.9298
- Recall: 0.9257
- F1: 0.9277
- Accuracy: 0.9812
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: 5e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- num_epochs: 20
Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|---|
| No log | 1.0 | 301 | 0.1139 | 0.8862 | 0.8862 | 0.8862 | 0.9694 |
| 0.2008 | 2.0 | 602 | 0.0925 | 0.8741 | 0.9155 | 0.8944 | 0.9729 |
| 0.2008 | 3.0 | 903 | 0.0910 | 0.8901 | 0.9125 | 0.9012 | 0.9747 |
| 0.0686 | 4.0 | 1204 | 0.1056 | 0.8947 | 0.9263 | 0.9102 | 0.9753 |
| 0.0501 | 5.0 | 1505 | 0.0921 | 0.9071 | 0.9305 | 0.9187 | 0.9775 |
| 0.0501 | 6.0 | 1806 | 0.0939 | 0.9062 | 0.9377 | 0.9217 | 0.9789 |
| 0.036 | 7.0 | 2107 | 0.1034 | 0.8926 | 0.9359 | 0.9137 | 0.9769 |
| 0.036 | 8.0 | 2408 | 0.1305 | 0.9019 | 0.9425 | 0.9218 | 0.9779 |
| 0.0219 | 9.0 | 2709 | 0.1320 | 0.9037 | 0.9335 | 0.9184 | 0.9778 |
| 0.0089 | 10.0 | 3010 | 0.1241 | 0.9271 | 0.9065 | 0.9167 | 0.9781 |
| 0.0089 | 11.0 | 3311 | 0.1386 | 0.9184 | 0.9311 | 0.9247 | 0.9791 |
| 0.0056 | 12.0 | 3612 | 0.1482 | 0.9094 | 0.9377 | 0.9233 | 0.9788 |
| 0.0056 | 13.0 | 3913 | 0.1550 | 0.9109 | 0.9311 | 0.9209 | 0.9783 |
| 0.0032 | 14.0 | 4214 | 0.1631 | 0.9078 | 0.9377 | 0.9225 | 0.9792 |
Framework versions
- Transformers 4.55.4
- Pytorch 2.8.0+cu126
- Datasets 4.0.0
- Tokenizers 0.21.4