wav2vec2-timit-demo
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: 0.4847
- Wer: 0.3462
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: 32
- 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 |
|---|---|---|---|---|
| 3.487 | 4.0 | 500 | 1.3466 | 1.0153 |
| 0.6134 | 8.0 | 1000 | 0.4807 | 0.4538 |
| 0.2214 | 12.0 | 1500 | 0.4684 | 0.3984 |
| 0.1233 | 16.0 | 2000 | 0.5070 | 0.3779 |
| 0.0847 | 20.0 | 2500 | 0.4965 | 0.3705 |
| 0.0611 | 24.0 | 3000 | 0.4881 | 0.3535 |
| 0.0464 | 28.0 | 3500 | 0.4847 | 0.3462 |
Framework versions
- Transformers 4.11.3
- Pytorch 1.10.2+cu102
- Datasets 1.18.3
- Tokenizers 0.10.3