openai/whisper-medium
This model is a fine-tuned version of openai/whisper-medium on the common_voice_11_0 dataset. It achieves the following results on the evaluation set:
- Loss: 1.1422
- Wer: 35.2207
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: 7000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.1137 | 4.02 | 1000 | 0.9072 | 40.0987 |
| 0.0153 | 9.02 | 2000 | 1.0351 | 38.7631 |
| 0.0042 | 14.01 | 3000 | 1.0507 | 36.4402 |
| 0.0013 | 19.0 | 4000 | 1.0924 | 36.2660 |
| 0.0003 | 23.02 | 5000 | 1.1422 | 35.2207 |
| 0.0001 | 28.02 | 6000 | 1.1688 | 35.3368 |
| 0.0001 | 33.01 | 7000 | 1.1768 | 35.5110 |
Framework versions
- Transformers 4.27.0.dev0
- Pytorch 1.13.1+cu117
- Datasets 2.9.1.dev0
- Tokenizers 0.13.2