Whisper Tiny it 6
This model is a fine-tuned version of openai/whisper-tiny on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.828768
- Wer: 46.277038
Model description
This model is the openai whisper small transformer adapted for Italian audio to text transcription. As part of the hyperparameter tuning process weight decay set to 0.1, attention dropout, encoder dropout and decoder dropout have been set to 0.1, the learning rate has been set to 1e-4, the number of decoder attention heads and encoder attention heads have been set to 8 however, it did not improved the performance on the evaluation set.
Intended uses & limitations
The model is available through its HuggingFace web app
Training and evaluation data
Data used for training is the initial 10% of train and validation of Italian Common Voice 11.0 from Mozilla Foundation. The dataset used for evaluation is the initial 10% of test of Italian Common Voice.
Training procedure
After loading the pre trained model, it has been trained on the dataset.
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-04
- 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: 4000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 1.7168 | 0.95 | 1000 | 1.2107 | 64.8087 |
| 1.1073 | 1.91 | 2000 | 0.9891 | 53.0019 |
| 1.3410 | 2.86 | 3000 | 0.8742 | 47.7676 |
| 0.6761 | 3.82 | 4000 | 0.8288 | 46.2770 |
Framework versions
- Transformers 4.26.0.dev0
- Pytorch 1.12.1+cu113
- Datasets 2.7.1
- Tokenizers 0.13.2
- Downloads last month
- 6