Whisper Tiny UZ - Bahriddin Muminov
This model is a fine-tuned version of openai/whisper-tiny on the Common Voice 16.1 dataset. It achieves the following results on the evaluation set:
- Loss: 0.5943
- Wer: 50.1992
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: 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: 1000
- training_steps: 10000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.6528 | 0.0352 | 2000 | 0.8591 | 62.8798 |
| 0.5201 | 0.0704 | 4000 | 0.7052 | 56.4951 |
| 0.4258 | 0.1056 | 6000 | 0.6407 | 53.7817 |
| 0.4136 | 0.1408 | 8000 | 0.6057 | 50.5588 |
| 0.4164 | 0.1760 | 10000 | 0.5943 | 50.1992 |
Framework versions
- Transformers 4.49.0
- Pytorch 2.6.0+cu124
- Datasets 3.4.1
- Tokenizers 0.21.1