Rajasthani-Magadhi-Marathi-Chhattisgarhi-Bhojpuri-Marwari-w2v-bert-2.0
This model is a fine-tuned version of facebook/w2v-bert-2.0 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 2.0250
- Cer: 0.1801
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: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- 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
- lr_scheduler_warmup_steps: 500
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 3.035 | 0.4383 | 300 | 1.1576 | 0.2526 |
| 1.0083 | 0.8766 | 600 | 0.9536 | 0.2142 |
| 0.7704 | 1.3141 | 900 | 0.8756 | 0.1959 |
| 0.6457 | 1.7524 | 1200 | 0.8496 | 0.1911 |
| 0.5242 | 2.1899 | 1500 | 0.8774 | 0.2010 |
| 0.437 | 2.6282 | 1800 | 0.9185 | 0.1866 |
| 0.351 | 3.0657 | 2100 | 1.0282 | 0.1906 |
| 0.271 | 3.5040 | 2400 | 1.0636 | 0.1867 |
| 0.2244 | 3.9423 | 2700 | 1.1005 | 0.1884 |
| 0.168 | 4.3798 | 3000 | 1.0549 | 0.1939 |
| 0.1331 | 4.8181 | 3300 | 1.1412 | 0.1867 |
| 0.0991 | 5.2557 | 3600 | 1.4552 | 0.1817 |
| 0.0773 | 5.6939 | 3900 | 1.4330 | 0.1816 |
| 0.0648 | 6.1315 | 4200 | 1.5016 | 0.1880 |
| 0.0467 | 6.5698 | 4500 | 1.6359 | 0.1813 |
| 0.0393 | 7.0073 | 4800 | 1.5973 | 0.1835 |
| 0.025 | 7.4456 | 5100 | 1.6842 | 0.1813 |
| 0.0217 | 7.8839 | 5400 | 1.8462 | 0.1801 |
| 0.0157 | 8.3214 | 5700 | 1.8816 | 0.1808 |
| 0.0112 | 8.7597 | 6000 | 1.9032 | 0.1808 |
| 0.0078 | 9.1972 | 6300 | 1.9836 | 0.1795 |
| 0.0059 | 9.6355 | 6600 | 2.0250 | 0.1801 |
Framework versions
- Transformers 4.57.6
- Pytorch 2.9.0+cu126
- Datasets 4.0.0
- Tokenizers 0.22.2