afrisenti-diacritized-scratch_scratch
This model is a fine-tuned version of roberta-base on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.1908
- Accuracy: 0.6437
- F1 Macro: 0.6442
- Composite Score: 0.6186
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: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 20
- mixed_precision_training: Native AMP
- label_smoothing_factor: 0.1
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 Macro | Composite Score |
|---|---|---|---|---|---|---|
| No log | 1.0 | 332 | 1.0888 | 0.3340 | 0.1669 | 0.1390 |
| 1.091 | 2.0 | 664 | 1.0788 | 0.4435 | 0.4038 | 0.3781 |
| 1.091 | 3.0 | 996 | 0.9375 | 0.5667 | 0.5638 | 0.5638 |
| 1.0167 | 4.0 | 1328 | 0.9561 | 0.5931 | 0.5929 | 0.5929 |
| 0.7803 | 5.0 | 1660 | 0.9845 | 0.6264 | 0.6294 | 0.6294 |
| 0.7803 | 6.0 | 1992 | 1.0312 | 0.6351 | 0.6370 | 0.6370 |
| 0.6012 | 7.0 | 2324 | 1.0842 | 0.6445 | 0.6441 | 0.6383 |
| 0.4942 | 8.0 | 2656 | 1.1326 | 0.6290 | 0.6256 | 0.6105 |
| 0.4942 | 9.0 | 2988 | 1.1550 | 0.6328 | 0.6316 | 0.6129 |
| 0.4446 | 10.0 | 3320 | 1.1908 | 0.6437 | 0.6442 | 0.6186 |
Framework versions
- Transformers 4.53.3
- Pytorch 2.6.0+cu124
- Datasets 4.4.1
- Tokenizers 0.21.2