Whisper Small Ru ORD 0.9 - Mizoru
This model is a fine-tuned version of openai/whisper-small on the ORD_0.9 dataset. It achieves the following results on the evaluation set:
- Loss: 1.0101
- Wer: 55.3480
- Cer: 30.6157
- Clean Wer: 49.0786
- Clean Cer: 25.0685
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: 16
- 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
- num_epochs: 4
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer | Clean Wer | Clean Cer |
|---|---|---|---|---|---|---|---|
| 1.0556 | 1.0 | 550 | 1.0824 | 56.6466 | 31.3266 | 49.6817 | 25.7468 |
| 0.979 | 2.0 | 1100 | 1.0150 | 55.8725 | 31.2389 | 49.7929 | 25.7459 |
| 0.8231 | 3.0 | 1650 | 1.0072 | 55.8588 | 30.5663 | 49.0675 | 24.9799 |
| 0.7372 | 4.0 | 2200 | 1.0101 | 55.3480 | 30.6157 | 49.0786 | 25.0685 |
Framework versions
- Transformers 4.41.0.dev0
- Pytorch 2.1.2
- Datasets 2.18.0
- Tokenizers 0.19.1
- Downloads last month
- 7