157-elevenlabs
This model is a fine-tuned version of openai/whisper-small on the 157-elevenlabs dataset. It achieves the following results on the evaluation set:
- Loss: 0.4340
- Wer: 25.4420
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: 8
- eval_batch_size: 4
- seed: 42
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| No log | 0 | 0 | 0.6890 | 41.0325 |
| 0.249 | 1.0 | 20 | 0.4580 | 27.8888 |
| 0.0763 | 2.0 | 40 | 0.4196 | 26.2618 |
| 0.0297 | 3.0 | 60 | 0.4145 | 26.3131 |
| 0.0163 | 4.0 | 80 | 0.4242 | 26.4284 |
| 0.0097 | 5.0 | 100 | 0.4272 | 26.5053 |
| 0.0064 | 6.0 | 120 | 0.4280 | 25.6213 |
| 0.0044 | 7.0 | 140 | 0.4317 | 25.4676 |
| 0.0035 | 8.0 | 160 | 0.4329 | 25.4035 |
| 0.003 | 9.0 | 180 | 0.4337 | 25.3907 |
| 0.0028 | 10.0 | 200 | 0.4340 | 25.4420 |
Framework versions
- Transformers 4.49.0
- Pytorch 2.5.1+cu121
- Datasets 3.4.1
- Tokenizers 0.21.0