wav2vec2-base-vi-vlsp2020-ft-vivos_phowhisper
This model is a fine-tuned version of minhvn1433/wav2vec2-base-vi-vlsp2020 on the MINHVN1433/VIVOS_PHOWHISPER - NA dataset. It achieves the following results on the evaluation set:
- Loss: 0.1322
- Wer: 0.0700
- Cer: 0.0289
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: 16
- eval_batch_size: 16
- seed: 42
- 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: 500.0
- training_steps: 5500
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 0.2192 | 0.7622 | 500 | 0.1563 | 0.0746 | 0.0315 |
| 0.1675 | 1.5244 | 1000 | 0.1435 | 0.0732 | 0.0307 |
| 0.1624 | 2.2866 | 1500 | 0.1394 | 0.0719 | 0.0300 |
| 0.1473 | 3.0488 | 2000 | 0.1365 | 0.0704 | 0.0294 |
| 0.1432 | 3.8110 | 2500 | 0.1356 | 0.0708 | 0.0295 |
| 0.1415 | 4.5732 | 3000 | 0.1345 | 0.0707 | 0.0293 |
| 0.1437 | 5.3354 | 3500 | 0.1344 | 0.0704 | 0.0290 |
| 0.1368 | 6.0976 | 4000 | 0.1336 | 0.0712 | 0.0292 |
| 0.1377 | 6.8598 | 4500 | 0.1321 | 0.0706 | 0.0292 |
| 0.1362 | 7.6220 | 5000 | 0.1335 | 0.0712 | 0.0291 |
| 0.1269 | 8.3841 | 5500 | 0.1322 | 0.0700 | 0.0289 |
Framework versions
- Transformers 5.12.1
- Pytorch 2.11.0+cu128
- Datasets 3.6.0
- Tokenizers 0.22.2