openai/whisper-medium
This model is a fine-tuned version of openai/whisper-medium on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.2589
- Wer: 10.2458
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: 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: 500
- training_steps: 4000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.2975 | 0.12 | 500 | 0.2211 | 11.1099 |
| 0.2856 | 1.07 | 1000 | 0.2027 | 10.5108 |
| 0.0727 | 2.01 | 1500 | 0.2071 | 10.1778 |
| 0.1546 | 2.13 | 2000 | 0.1978 | 10.5596 |
| 0.0166 | 3.08 | 2500 | 0.2328 | 9.9899 |
| 0.0076 | 4.02 | 3000 | 0.2436 | 10.3463 |
| 0.0042 | 4.14 | 3500 | 0.2497 | 10.5311 |
| 0.0066 | 5.09 | 4000 | 0.2589 | 10.2458 |
Framework versions
- Transformers 4.27.0.dev0
- Pytorch 1.13.1+cu117
- Datasets 2.9.1.dev0
- Tokenizers 0.13.2