whisper-small-Yfreq_speed_pause
This model is a fine-tuned version of openai/whisper-small on the aihub old50 adult pause changed dataset. It achieves the following results on the evaluation set:
- Loss: 0.1711
- Cer: 7.3719
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: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 50
- num_epochs: 2
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 0.2956 | 0.1289 | 100 | 0.2468 | 6.1325 |
| 0.1584 | 0.2579 | 200 | 0.2170 | 5.7213 |
| 0.1965 | 0.3868 | 300 | 0.2103 | 5.9211 |
| 0.1558 | 0.5158 | 400 | 0.2022 | 5.3043 |
| 0.1337 | 0.6447 | 500 | 0.1980 | 5.4511 |
| 0.1174 | 0.7737 | 600 | 0.1854 | 5.2808 |
| 0.1183 | 0.9026 | 700 | 0.1809 | 5.0341 |
| 0.0691 | 1.0316 | 800 | 0.1765 | 5.0752 |
| 0.0561 | 1.1605 | 900 | 0.1799 | 5.0047 |
| 0.0547 | 1.2895 | 1000 | 0.1765 | 4.9460 |
| 0.0523 | 1.4184 | 1100 | 0.1765 | 6.2324 |
| 0.0511 | 1.5474 | 1200 | 0.1742 | 6.5848 |
| 0.0512 | 1.6763 | 1300 | 0.1729 | 8.1003 |
| 0.0452 | 1.8053 | 1400 | 0.1715 | 6.7199 |
| 0.0465 | 1.9342 | 1500 | 0.1711 | 7.3719 |
Framework versions
- Transformers 4.46.0.dev0
- Pytorch 2.4.1+cu121
- Datasets 3.0.1
- Tokenizers 0.20.1