whisper-medium-combined-lin-sna-lug-olusola
This model is a fine-tuned version of openai/whisper-medium on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.5651
- Wer: 0.3436
- Cer: 0.1221
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: 0.0001
- train_batch_size: 8
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 32
- 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: 500
- num_epochs: 30
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 2.5540 | 0.5021 | 500 | 0.6311 | 0.4609 | 0.1696 |
| 2.0345 | 1.0040 | 1000 | 0.5528 | 0.3869 | 0.1481 |
| 1.7164 | 1.5062 | 1500 | 0.5335 | 0.4307 | 0.1922 |
| 1.3803 | 2.0080 | 2000 | 0.5164 | 0.4613 | 0.1879 |
| 1.2623 | 2.5102 | 2500 | 0.5028 | 0.3283 | 0.1112 |
| 0.9971 | 3.0121 | 3000 | 0.5225 | 0.3535 | 0.1288 |
| 0.8955 | 3.5142 | 3500 | 0.5332 | 0.3375 | 0.1205 |
| 0.7047 | 4.0161 | 4000 | 0.5651 | 0.3436 | 0.1221 |
Framework versions
- Transformers 5.12.1
- Pytorch 2.11.0+cu128
- Datasets 4.0.0
- Tokenizers 0.22.2