wav2vec2-base-vi-vlsp2020-ft-ViMD_south_phowhisper
This model is a fine-tuned version of minhvn1433/wav2vec2-base-vi-vlsp2020 on the MINHVN1433/VIMD_SOUTH_PHOWHISPER - NA dataset. It achieves the following results on the evaluation set:
- Loss: 0.4545
- Wer: 0.1702
- Cer: 0.1048
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: 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: 3000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 0.3712 | 3.5714 | 500 | 0.4253 | 0.1751 | 0.1063 |
| 0.3271 | 7.1429 | 1000 | 0.4239 | 0.1714 | 0.1043 |
| 0.3004 | 10.7143 | 1500 | 0.4486 | 0.1705 | 0.1046 |
| 0.2791 | 14.2857 | 2000 | 0.4667 | 0.1728 | 0.1054 |
| 0.2688 | 17.8571 | 2500 | 0.4545 | 0.1702 | 0.1048 |
| 0.2664 | 21.4286 | 3000 | 0.4709 | 0.1702 | 0.1047 |
Framework versions
- Transformers 5.13.0
- Pytorch 2.11.0+cu128
- Datasets 3.6.0
- Tokenizers 0.22.2