whispertiny_kokoro_styletts_25sil_v1_1
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: 0.1156
- Wer: 16.2812
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 |
|---|---|---|---|---|
| 0.109 | 0.2286 | 1000 | 0.1684 | 34.3298 |
| 0.0901 | 0.4571 | 2000 | 0.1226 | 22.9763 |
| 0.0517 | 0.6857 | 3000 | 0.1073 | 19.1590 |
| 0.0437 | 0.9143 | 4000 | 0.1197 | 20.4534 |
| 0.0282 | 1.1429 | 5000 | 0.1101 | 17.6807 |
| 0.0175 | 1.3714 | 6000 | 0.1061 | 15.9264 |
| 0.0141 | 1.6 | 7000 | 0.1020 | 15.7753 |
| 0.0218 | 1.8286 | 8000 | 0.1095 | 16.3535 |
| 0.0143 | 2.0571 | 9000 | 0.1160 | 16.6360 |
| 0.015 | 2.2857 | 10000 | 0.1156 | 16.2812 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.3.0+cu121
- Datasets 4.0.0
- Tokenizers 0.19.1