combined-diacritized-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: 1.1767
- Accuracy: 0.6415
- F1 Macro: 0.6382
- Composite Score: 0.6350
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: 3e-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: 0.2
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 Macro | Composite Score |
|---|---|---|---|---|---|---|
| 1.0733 | 1.0 | 1059 | 1.0975 | 0.3989 | 0.3155 | 0.2977 |
| 0.8713 | 2.0 | 2118 | 1.0044 | 0.5410 | 0.5456 | 0.5456 |
| 0.799 | 3.0 | 3177 | 1.0575 | 0.5421 | 0.5455 | 0.5455 |
| 0.7059 | 4.0 | 4236 | 1.0504 | 0.6181 | 0.6203 | 0.6203 |
| 0.6317 | 5.0 | 5295 | 1.0562 | 0.6403 | 0.6418 | 0.6418 |
| 0.5989 | 6.0 | 6354 | 1.0962 | 0.6332 | 0.6315 | 0.6315 |
| 0.5717 | 7.0 | 7413 | 1.1273 | 0.6328 | 0.6313 | 0.6313 |
| 0.5593 | 8.0 | 8472 | 1.1278 | 0.6471 | 0.6463 | 0.6463 |
| 0.5459 | 9.0 | 9531 | 1.1632 | 0.6475 | 0.6458 | 0.6458 |
| 0.5373 | 10.0 | 10590 | 1.1553 | 0.6460 | 0.6451 | 0.6451 |
| 0.5274 | 11.0 | 11649 | 1.1767 | 0.6415 | 0.6382 | 0.6350 |
Framework versions
- Transformers 4.53.3
- Pytorch 2.6.0+cu124
- Datasets 4.4.1
- Tokenizers 0.21.2