whispertiny_styletts_25sil_v1_0
This model is a fine-tuned version of openai/whisper-tiny on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.5157
- Wer: 38.8896
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: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 10000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0638 | 0.2286 | 1000 | 1.1389 | 45.5519 |
| 0.0383 | 0.4571 | 2000 | 1.1848 | 42.2339 |
| 0.029 | 0.6857 | 3000 | 1.2547 | 40.8870 |
| 0.0184 | 0.9143 | 4000 | 1.3745 | 40.8870 |
| 0.0177 | 1.1429 | 5000 | 1.3901 | 39.3561 |
| 0.0171 | 1.3714 | 6000 | 1.3677 | 39.4809 |
| 0.0107 | 1.6 | 7000 | 1.4280 | 38.8239 |
| 0.0137 | 1.8286 | 8000 | 1.4553 | 39.5401 |
| 0.0095 | 2.0571 | 9000 | 1.4822 | 39.4350 |
| 0.0043 | 2.2857 | 10000 | 1.5157 | 38.8896 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.3.0+cu121
- Datasets 4.0.0
- Tokenizers 0.19.1