wav2vec2-base-20sec-timit-and-dementiabank
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: 0.4338
- Wer: 0.2313
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: 4
- 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: 20
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 3.6839 | 2.53 | 500 | 2.7287 | 1.0 |
| 0.8708 | 5.05 | 1000 | 0.5004 | 0.3490 |
| 0.2879 | 7.58 | 1500 | 0.4411 | 0.2872 |
| 0.1877 | 10.1 | 2000 | 0.4359 | 0.2594 |
| 0.1617 | 12.63 | 2500 | 0.4404 | 0.2492 |
| 0.1295 | 15.15 | 3000 | 0.4356 | 0.2418 |
| 0.1146 | 17.68 | 3500 | 0.4338 | 0.2313 |
Framework versions
- Transformers 4.11.3
- Pytorch 1.10.0+cu111
- Datasets 1.18.3
- Tokenizers 0.10.3