IoanaLivia/real-voices-youtube-whisper-small
This model is a fine-tuned version of openai/whisper-small on the IoanaLivia/real-voices-youtube-horoscope dataset. It achieves the following results on the evaluation set:
- Loss: 0.3917
- Wer: 19.4258
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: 8
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- num_epochs: 40
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| No log | 0 | 0 | 0.4170 | 30.1014 |
| 0.4264 | 1.0 | 24 | 0.3082 | 21.4372 |
| 0.1957 | 2.0 | 48 | 0.2799 | 20.4917 |
| 0.1032 | 3.0 | 72 | 0.2904 | 20.3369 |
| 0.0535 | 4.0 | 96 | 0.3016 | 20.2338 |
| 0.0261 | 5.0 | 120 | 0.3226 | 20.5604 |
| 0.0148 | 6.0 | 144 | 0.3466 | 20.7495 |
| 0.0091 | 7.0 | 168 | 0.3596 | 21.1793 |
| 0.0063 | 8.0 | 192 | 0.3696 | 20.9386 |
| 0.0052 | 9.0 | 216 | 0.3703 | 19.7009 |
| 0.004 | 10.0 | 240 | 0.3749 | 19.8212 |
| 0.0032 | 11.0 | 264 | 0.3846 | 20.1822 |
| 0.0027 | 12.0 | 288 | 0.3867 | 19.5462 |
| 0.0023 | 13.0 | 312 | 0.3917 | 19.4258 |
| 0.0021 | 14.0 | 336 | 0.3948 | 20.6120 |
| 0.0019 | 15.0 | 360 | 0.3980 | 19.5633 |
| 0.0017 | 16.0 | 384 | 0.4014 | 19.6321 |
| 0.0016 | 17.0 | 408 | 0.4038 | 19.4946 |
| 0.0015 | 18.0 | 432 | 0.4067 | 19.4946 |
| 0.0014 | 19.0 | 456 | 0.4088 | 19.4774 |
| 0.0013 | 20.0 | 480 | 0.4113 | 19.6321 |
| 0.0013 | 21.0 | 504 | 0.4133 | 19.5805 |
| 0.0012 | 22.0 | 528 | 0.4153 | 19.6493 |
| 0.0011 | 23.0 | 552 | 0.4173 | 19.6493 |
| 0.0011 | 24.0 | 576 | 0.4189 | 19.6493 |
| 0.001 | 25.0 | 600 | 0.4203 | 19.7009 |
| 0.001 | 26.0 | 624 | 0.4219 | 19.7181 |
| 0.001 | 27.0 | 648 | 0.4232 | 19.7353 |
| 0.0009 | 28.0 | 672 | 0.4245 | 19.8040 |
| 0.0009 | 29.0 | 696 | 0.4257 | 19.8384 |
| 0.0009 | 30.0 | 720 | 0.4268 | 19.8900 |
| 0.0009 | 31.0 | 744 | 0.4276 | 19.8556 |
| 0.0009 | 32.0 | 768 | 0.4286 | 19.8556 |
| 0.0008 | 33.0 | 792 | 0.4293 | 19.7868 |
| 0.0008 | 34.0 | 816 | 0.4300 | 19.8728 |
| 0.0008 | 35.0 | 840 | 0.4307 | 19.7696 |
| 0.0008 | 36.0 | 864 | 0.4311 | 19.8728 |
| 0.0008 | 37.0 | 888 | 0.4315 | 19.8212 |
| 0.0008 | 38.0 | 912 | 0.4317 | 19.8556 |
| 0.0008 | 39.0 | 936 | 0.4319 | 19.9244 |
| 0.0008 | 40.0 | 960 | 0.4319 | 19.8212 |
Framework versions
- Transformers 4.51.3
- Pytorch 2.6.0+cu124
- Datasets 3.6.0
- Tokenizers 0.21.1