Whisper Base Hu v5 - cleaned
This model is a fine-tuned version of openai/whisper-tiny on the Common Voice 16.1 hu cleaned dataset. It achieves the following results on the evaluation set:
- Loss: 0.1705
- Wer Ortho: 16.1247
- Wer: 15.1778
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: 2.5e-05
- train_batch_size: 32
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant_with_warmup
- lr_scheduler_warmup_steps: 300
- training_steps: 10000
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer Ortho | Wer |
---|---|---|---|---|---|
0.2084 | 0.83 | 1000 | 0.2183 | 26.0872 | 24.6511 |
0.0992 | 1.66 | 2000 | 0.1716 | 20.5220 | 19.2263 |
0.0443 | 2.49 | 3000 | 0.1545 | 18.3604 | 17.3452 |
0.0242 | 3.32 | 4000 | 0.1563 | 17.7216 | 16.6602 |
0.0127 | 4.15 | 5000 | 0.1551 | 17.1216 | 16.1489 |
0.0173 | 4.98 | 6000 | 0.1584 | 17.3087 | 16.3194 |
0.0111 | 5.81 | 7000 | 0.1670 | 16.9119 | 15.8338 |
0.0087 | 6.64 | 8000 | 0.1653 | 17.1087 | 16.0428 |
0.0059 | 7.48 | 9000 | 0.1669 | 16.5344 | 15.5219 |
0.007 | 8.31 | 10000 | 0.1705 | 16.1247 | 15.1778 |
Framework versions
- Transformers 4.37.2
- Pytorch 2.1.0+cu121
- Datasets 2.17.0
- Tokenizers 0.15.1
- Downloads last month
- 4