UrduNER
This model is a fine-tuned version of urduhack/roberta-urdu-small on the wikiann dataset. It achieves the following results on the evaluation set:
- Loss: 0.1163
- Overall Precision: 0.9540
- Overall Recall: 0.9553
- Overall F1: 0.9546
- Overall Accuracy: 0.9836
- Loc F1: 0.9643
- Org F1: 0.9448
- Per F1: 0.9491
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: linear
- num_epochs: 7
Training results
| Training Loss | Epoch | Step | Validation Loss | Overall Precision | Overall Recall | Overall F1 | Overall Accuracy | Loc F1 | Org F1 | Per F1 |
|---|---|---|---|---|---|---|---|---|---|---|
| 0.248 | 1.0 | 1250 | 0.0920 | 0.8906 | 0.8991 | 0.8948 | 0.9687 | 0.9086 | 0.8686 | 0.8995 |
| 0.1169 | 2.0 | 2500 | 0.0761 | 0.9302 | 0.9390 | 0.9346 | 0.9791 | 0.9501 | 0.9045 | 0.9400 |
| 0.07 | 3.0 | 3750 | 0.0831 | 0.9394 | 0.9451 | 0.9422 | 0.9805 | 0.9505 | 0.9348 | 0.9361 |
| 0.029 | 4.0 | 5000 | 0.1102 | 0.9311 | 0.9431 | 0.9371 | 0.9784 | 0.9469 | 0.9305 | 0.9279 |
| 0.0134 | 5.0 | 6250 | 0.1225 | 0.9442 | 0.9519 | 0.9480 | 0.9820 | 0.9593 | 0.9438 | 0.9337 |
| 0.0107 | 6.0 | 7500 | 0.1087 | 0.9515 | 0.9566 | 0.9541 | 0.9837 | 0.9660 | 0.9423 | 0.9466 |
| 0.005 | 7.0 | 8750 | 0.1163 | 0.9540 | 0.9553 | 0.9546 | 0.9836 | 0.9643 | 0.9448 | 0.9491 |
Framework versions
- Transformers 4.32.0
- Pytorch 2.0.1+cu118
- Datasets 2.14.4
- Tokenizers 0.13.3