whisper-small-chinese-tw-minnan
This model is a fine-tuned version of openai/whisper-medium on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 2.1011
- Wer: 253.6647
- Cer: 194.5002
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: 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: 500
- training_steps: 5000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 0.3001 | 2.3256 | 1000 | 1.6189 | 180.3509 | 109.0310 |
| 0.069 | 4.6512 | 2000 | 1.7769 | 390.2924 | 270.4935 |
| 0.0218 | 6.9767 | 3000 | 1.8940 | 195.6530 | 131.7026 |
| 0.0072 | 9.3023 | 4000 | 2.0372 | 213.8596 | 143.6605 |
| 0.005 | 11.6279 | 5000 | 2.1011 | 253.6647 | 194.5002 |
Framework versions
- Transformers 4.50.3
- Pytorch 2.6.0+cu124
- Datasets 3.5.0
- Tokenizers 0.21.1