wav2vec_trained
This model is a fine-tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.0337
- Wer: 0.1042
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: 0.0001
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 4.3849 | 2.21 | 500 | 2.9148 | 1.0 |
| 1.9118 | 4.42 | 1000 | 0.9627 | 0.5833 |
| 0.7596 | 6.64 | 1500 | 0.8953 | 0.3542 |
| 0.4602 | 8.85 | 2000 | 0.3325 | 0.2083 |
| 0.331 | 11.06 | 2500 | 0.3084 | 0.2083 |
| 0.2474 | 13.27 | 3000 | 0.0960 | 0.1667 |
| 0.1934 | 15.49 | 3500 | 0.1276 | 0.125 |
| 0.156 | 17.7 | 4000 | 0.0605 | 0.0833 |
| 0.1244 | 19.91 | 4500 | 0.0831 | 0.1458 |
| 0.1006 | 22.12 | 5000 | 0.0560 | 0.125 |
| 0.0827 | 24.34 | 5500 | 0.0395 | 0.0833 |
| 0.0723 | 26.55 | 6000 | 0.0573 | 0.0833 |
| 0.0606 | 28.76 | 6500 | 0.0337 | 0.1042 |
Framework versions
- Transformers 4.17.0
- Pytorch 1.11.0+cu113
- Datasets 1.18.3
- Tokenizers 0.12.1