whisper-base-turkish-1.1
This model is a fine-tuned version of ysdede/whisper-base-turkish-1 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.3646
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: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- training_steps: 25000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.2878 | 0.1 | 2500 | 0.4129 |
| 0.2732 | 0.2 | 5000 | 0.4008 |
| 0.2799 | 0.3 | 7500 | 0.3906 |
| 0.244 | 0.4 | 10000 | 0.3819 |
| 0.2304 | 0.5 | 12500 | 0.3784 |
| 0.2318 | 0.6 | 15000 | 0.3707 |
| 0.2245 | 0.7 | 17500 | 0.3675 |
| 0.214 | 0.8 | 20000 | 0.3648 |
| 0.2026 | 0.9 | 22500 | 0.3640 |
| 0.1943 | 1.0 | 25000 | 0.3646 |
Framework versions
- Transformers 4.45.2
- Pytorch 2.5.1+cu124
- Datasets 3.2.0
- Tokenizers 0.20.3