Whisper Small ks - Muneeb Haroon
This model is a fine-tuned version of openai/whisper-small on the whisper-kashmiri dataset. It achieves the following results on the evaluation set:
- Loss: 1.1578
- Wer: 39.8077
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: 2
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 8
- optimizer: Use OptimizerNames.ADAMW_TORCH 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: 500
- training_steps: 10000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0123 | 21.2811 | 1000 | 0.9382 | 48.125 |
| 0.0051 | 42.5622 | 2000 | 0.9946 | 42.4519 |
| 0.0032 | 63.8432 | 3000 | 1.0278 | 41.3942 |
| 0.0 | 85.1081 | 4000 | 1.1138 | 40.5288 |
| 0.0 | 106.3892 | 5000 | 1.1578 | 39.8077 |
| 0.0 | 127.6703 | 6000 | 1.1869 | 39.8077 |
| 0.0 | 148.9514 | 7000 | 1.2211 | 40.0 |
| 0.0 | 170.2162 | 8000 | 1.2430 | 40.2404 |
| 0.0 | 191.4973 | 9000 | 1.2679 | 40.2885 |
| 0.0 | 212.7784 | 10000 | 1.2762 | 40.3365 |
Framework versions
- Transformers 4.49.0
- Pytorch 2.5.1+cu124
- Datasets 3.2.0
- Tokenizers 0.21.0