whisper-small-0DB-3r6
This model is a fine-tuned version of openai/whisper-small on the None dataset. It achieves the following results on the evaluation set:
- Cer: 38.6899
- Loss: 0.4287
- Wer: 84.8672
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: 3e-06
- train_batch_size: 32
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 100
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Cer | Validation Loss | Wer |
|---|---|---|---|---|---|
| 4.6407 | 1.2675 | 100 | 71.2115 | 1.1388 | 100.1235 |
| 2.9213 | 2.5350 | 200 | 45.1097 | 0.7472 | 92.2792 |
| 2.2742 | 3.8025 | 300 | 40.0584 | 0.6162 | 88.2644 |
| 1.8082 | 5.0637 | 400 | 40.1791 | 0.5245 | 85.2378 |
| 1.3140 | 6.3312 | 500 | 40.0885 | 0.4390 | 85.4849 |
| 1.1840 | 7.5987 | 600 | 39.2131 | 0.4320 | 85.0525 |
| 1.0385 | 8.8662 | 700 | 38.6899 | 0.4287 | 84.8672 |
Framework versions
- Transformers 5.0.0
- Pytorch 2.10.0+cu128
- Datasets 4.8.3
- Tokenizers 0.22.2