whisper-base-ft-common-language-id
This model is a fine-tuned version of openai/whisper-base on the common_language dataset. It achieves the following results on the evaluation set:
- Loss: 1.0725
- Accuracy: 0.7525
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: 32
- eval_batch_size: 16
- seed: 0
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 10.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 2.5291 | 1.0 | 694 | 2.4787 | 0.4806 |
| 1.5801 | 2.0 | 1388 | 1.6258 | 0.6260 |
| 1.0144 | 3.0 | 2082 | 1.2886 | 0.6816 |
| 0.7442 | 4.0 | 2776 | 1.0783 | 0.7237 |
| 0.4802 | 5.0 | 3470 | 1.0582 | 0.7266 |
| 0.3378 | 6.0 | 4164 | 1.0173 | 0.7417 |
| 0.1941 | 7.0 | 4858 | 1.0054 | 0.7446 |
| 0.1424 | 8.0 | 5552 | 1.0213 | 0.7508 |
| 0.1242 | 9.0 | 6246 | 1.0567 | 0.7495 |
| 0.1527 | 10.0 | 6940 | 1.0725 | 0.7525 |
Framework versions
- Transformers 4.27.0.dev0
- Pytorch 1.13.1
- Datasets 2.9.0
- Tokenizers 0.13.2