metadata
license: apache-2.0
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: whisper-small-nl
results: []
whisper-small-nl
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: 0.3777
- Wer: 19.8814
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: 512
- eval_batch_size: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 50
- training_steps: 1000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.8378 | 0.1 | 100 | 0.4933 | 23.8827 |
| 0.5547 | 0.2 | 200 | 0.4476 | 21.0578 |
| 0.3905 | 0.3 | 300 | 0.4335 | 21.1689 |
| 0.3766 | 0.4 | 400 | 0.4267 | 20.0528 |
| 0.4164 | 0.5 | 500 | 0.4139 | 21.4329 |
| 0.2939 | 0.6 | 600 | 0.3864 | 18.3671 |
| 0.2632 | 0.7 | 700 | 0.3864 | 18.4319 |
| 0.6066 | 0.8 | 800 | 0.3804 | 19.2748 |
| 0.2075 | 1.09 | 900 | 0.3794 | 18.8904 |
| 0.2102 | 1.19 | 1000 | 0.3777 | 19.8814 |
Framework versions
- Transformers 4.26.0.dev0
- Pytorch 1.13.0+cu117
- Datasets 2.7.1.dev0
- Tokenizers 0.13.2