distil-whisper-large-v2-pt
This model is a fine-tuned version of distil-whisper/distil-large-v2 on the mozilla-foundation/common_voice_13_0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.3028
- Wer Ortho: 0.1649
- Wer: 0.1104
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: 7e-06
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant_with_warmup
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 3
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer Ortho | Wer |
|---|---|---|---|---|---|
| 1.6148 | 0.5 | 900 | 0.4448 | 0.2227 | 0.1690 |
| 0.3709 | 0.99 | 1800 | 0.3524 | 0.1927 | 0.1367 |
| 0.2619 | 1.49 | 2700 | 0.3266 | 0.1751 | 0.1213 |
| 0.2143 | 1.98 | 3600 | 0.3085 | 0.1726 | 0.1168 |
| 0.1219 | 2.48 | 4500 | 0.3070 | 0.1639 | 0.1112 |
| 0.1256 | 2.98 | 5400 | 0.3028 | 0.1649 | 0.1104 |
Framework versions
- Transformers 4.35.0
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1