Whisper Base Indonesian
This model is a fine-tuned version of arun100/whisper-base-id-1 on the google/fleurs id_id dataset. It achieves the following results on the evaluation set:
- Loss: 0.5254
- Wer: 26.7478
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: 5e-07
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- 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 |
|---|---|---|---|---|
| 0.3799 | 45.0 | 500 | 0.5510 | 28.5103 |
| 0.1679 | 90.0 | 1000 | 0.5254 | 26.7478 |
| 0.0785 | 136.0 | 1500 | 0.5336 | 27.1386 |
| 0.0408 | 181.0 | 2000 | 0.5439 | 27.1755 |
| 0.0266 | 227.0 | 2500 | 0.5513 | 27.0354 |
| 0.02 | 272.0 | 3000 | 0.5569 | 28.0826 |
| 0.0159 | 318.0 | 3500 | 0.5612 | 28.5767 |
| 0.0136 | 363.0 | 4000 | 0.5645 | 30.1254 |
| 0.0124 | 409.0 | 4500 | 0.5667 | 28.7611 |
| 0.012 | 454.0 | 5000 | 0.5674 | 28.7021 |
Framework versions
- Transformers 4.37.0.dev0
- Pytorch 2.1.2+cu121
- Datasets 2.16.2.dev0
- Tokenizers 0.15.0