whisper-small-santali-sanlish-1934-lr1e5
This model is a fine-tuned version of openai/whisper-small on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.7348
- Wer: 33.4149
- Cer: 7.5411
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: 1e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- 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: 153
- num_epochs: 25
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 4.1433 | 1.0 | 99 | 0.8546 | 55.5692 | 13.8937 |
| 1.2416 | 2.0 | 198 | 0.5976 | 42.4725 | 10.3068 |
| 0.5530 | 3.0 | 297 | 0.5570 | 40.2693 | 9.3129 |
| 0.2314 | 4.0 | 396 | 0.6078 | 40.0245 | 8.8807 |
| 0.1007 | 5.0 | 495 | 0.6054 | 35.4957 | 8.2325 |
| 0.0608 | 6.0 | 594 | 0.6251 | 35.3733 | 8.1245 |
| 0.0368 | 7.0 | 693 | 0.6465 | 36.2301 | 8.2973 |
| 0.0208 | 8.0 | 792 | 0.6840 | 35.2509 | 8.2757 |
| 0.0110 | 9.0 | 891 | 0.6970 | 34.8837 | 8.0164 |
| 0.0068 | 10.0 | 990 | 0.6951 | 33.7821 | 7.6491 |
| 0.0029 | 11.0 | 1089 | 0.7220 | 35.3733 | 7.9732 |
| 0.0025 | 12.0 | 1188 | 0.7137 | 33.4149 | 7.5194 |
| 0.0016 | 13.0 | 1287 | 0.7023 | 33.5373 | 7.5411 |
| 0.0025 | 14.0 | 1386 | 0.7017 | 34.0269 | 7.8003 |
| 0.0012 | 15.0 | 1485 | 0.7127 | 33.6597 | 7.6059 |
| 0.0007 | 16.0 | 1584 | 0.7190 | 33.9045 | 7.6923 |
| 0.0006 | 17.0 | 1683 | 0.7228 | 33.7821 | 7.5843 |
| 0.0006 | 18.0 | 1782 | 0.7258 | 33.6597 | 7.5627 |
| 0.0005 | 19.0 | 1881 | 0.7285 | 33.6597 | 7.5627 |
| 0.0005 | 20.0 | 1980 | 0.7313 | 33.5373 | 7.5843 |
| 0.0005 | 21.0 | 2079 | 0.7333 | 33.5373 | 7.5627 |
| 0.0005 | 22.0 | 2178 | 0.7348 | 33.4149 | 7.5411 |
Framework versions
- Transformers 5.15.0
- Pytorch 2.11.0+cu128
- Datasets 4.0.0
- Tokenizers 0.22.2