Whisper large-v2 Test
This model is a fine-tuned version of openai/whisper-large-v2 on the Common Voice 16 Albanian dataset. It achieves the following results on the evaluation set:
- Loss: 0.7073
- Wer: 34.0530
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: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 5000
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.1135 | 4.63 | 500 | 0.6519 | 44.8880 |
| 0.02 | 9.26 | 1000 | 0.6575 | 39.3483 |
| 0.0075 | 13.89 | 1500 | 0.6073 | 35.6823 |
| 0.0016 | 18.52 | 2000 | 0.6347 | 34.9084 |
| 0.0008 | 23.15 | 2500 | 0.6484 | 34.9491 |
| 0.0001 | 27.78 | 3000 | 0.6765 | 34.4196 |
| 0.0001 | 32.41 | 3500 | 0.6897 | 33.9308 |
| 0.0001 | 37.04 | 4000 | 0.6988 | 34.1752 |
| 0.0001 | 41.67 | 4500 | 0.7048 | 33.9715 |
| 0.0001 | 46.3 | 5000 | 0.7073 | 34.0530 |
Framework versions
- Transformers 4.39.0.dev0
- Pytorch 2.1.0+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2