whisper-small-cer
This model is a fine-tuned version of openai/whisper-small on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.0002
- Cer: 0.0
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-06
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 5
- training_steps: 400
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Cer |
---|---|---|---|---|
3.6274 | 0.8 | 20 | 2.6626 | 2.0787 |
1.8902 | 1.6 | 40 | 1.2100 | 1.6333 |
0.6287 | 2.4 | 60 | 0.0808 | 0.9899 |
0.0193 | 3.2 | 80 | 0.0118 | 0.9404 |
0.0022 | 4.0 | 100 | 0.0022 | 0.2970 |
0.0006 | 4.8 | 120 | 0.0007 | 0.0 |
0.0005 | 5.6 | 140 | 0.0005 | 0.0 |
0.0004 | 6.4 | 160 | 0.0004 | 0.0 |
0.0003 | 7.2 | 180 | 0.0004 | 0.0 |
0.0003 | 8.0 | 200 | 0.0004 | 0.0 |
0.0003 | 8.8 | 220 | 0.0003 | 0.0 |
0.0002 | 9.6 | 240 | 0.0003 | 0.0 |
0.0002 | 10.4 | 260 | 0.0003 | 0.0 |
0.0002 | 11.2 | 280 | 0.0003 | 0.0 |
0.0002 | 12.0 | 300 | 0.0003 | 0.0 |
0.0002 | 12.8 | 320 | 0.0002 | 0.0 |
0.0002 | 13.6 | 340 | 0.0002 | 0.0 |
0.0002 | 14.4 | 360 | 0.0002 | 0.0 |
0.0002 | 15.2 | 380 | 0.0002 | 0.0 |
0.0002 | 16.0 | 400 | 0.0002 | 0.0 |
Framework versions
- Transformers 4.39.3
- Pytorch 2.2.2+cu121
- Datasets 2.14.5
- Tokenizers 0.15.2
- Downloads last month
- 2