wav2vec2-base-vi-vlsp2020-ft-VietMed_labeled_phowhisper
This model is a fine-tuned version of minhvn1433/wav2vec2-base-vi-vlsp2020 on the MINHVN1433/VIETMED_LABELED_PHOWHISPER - NA dataset. It achieves the following results on the evaluation set:
- Loss: 1.3017
- Wer: 0.2629
- Cer: 0.2034
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 |
|---|---|---|---|---|---|
| 0.5332 | 2.7933 | 500 | 1.3017 | 0.2629 | 0.2034 |
| 0.4427 | 5.5866 | 1000 | 1.2293 | 0.2648 | 0.2042 |
| 0.3902 | 8.3799 | 1500 | 1.2759 | 0.2674 | 0.2055 |
| 0.4016 | 11.1732 | 2000 | 1.2788 | 0.2638 | 0.2031 |
| 0.3860 | 13.9665 | 2500 | 1.2983 | 0.2661 | 0.2046 |
| 0.3862 | 16.7598 | 3000 | 1.3498 | 0.2650 | 0.2045 |
| 0.3706 | 19.5531 | 3500 | 1.3760 | 0.2660 | 0.2044 |
| 0.3599 | 22.3464 | 4000 | 1.3890 | 0.2653 | 0.2047 |
| 0.3562 | 25.1397 | 4500 | 1.4151 | 0.2685 | 0.2057 |
| 0.3154 | 27.9330 | 5000 | 1.4182 | 0.2677 | 0.2056 |
| 0.3459 | 30.7263 | 5500 | 1.3970 | 0.2658 | 0.2053 |
Framework versions
- Transformers 5.12.1
- Pytorch 2.11.0+cu128
- Datasets 3.6.0
- Tokenizers 0.22.2