whispertiny_kokoro_styletts_25sil_v1_3
This model is a fine-tuned version of openai/whisper-tiny on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.1001
- Wer: 7.5314
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
- training_steps: 10000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0972 | 0.2258 | 1000 | 0.1600 | 16.4139 |
| 0.0523 | 0.4516 | 2000 | 0.1191 | 11.9727 |
| 0.0428 | 0.6774 | 3000 | 0.1046 | 9.9245 |
| 0.0363 | 0.9031 | 4000 | 0.0994 | 9.0765 |
| 0.0222 | 1.1289 | 5000 | 0.1024 | 8.8897 |
| 0.0218 | 1.3547 | 6000 | 0.1017 | 7.7542 |
| 0.027 | 1.5805 | 7000 | 0.0976 | 7.8261 |
| 0.0153 | 1.8063 | 8000 | 0.0956 | 7.3805 |
| 0.0116 | 2.0321 | 9000 | 0.0950 | 7.1577 |
| 0.0191 | 2.2578 | 10000 | 0.1001 | 7.5314 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.3.0+cu121
- Datasets 4.0.0
- Tokenizers 0.19.1