whisper-large-v3-turbo-finetuned-appen-dhc
This model is a fine-tuned version of openai/whisper-large-v3-turbo on the whisper-large-finetune-appen-dhc dataset. It achieves the following results on the evaluation set:
- Loss: 0.8896
- Model Preparation Time: 0.0074
- Wer: 0.4650
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: 0.0003
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 64
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 400
- num_epochs: 8
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Model Preparation Time | Wer |
|---|---|---|---|---|---|
| 0.5406 | 1.0 | 250 | 0.8937 | 0.0074 | 0.5801 |
| 0.6665 | 2.0 | 500 | 0.8749 | 0.0074 | 0.4803 |
| 0.5642 | 3.0 | 750 | 0.6730 | 0.0074 | 0.3495 |
| 0.373 | 4.0 | 1000 | 0.5504 | 0.0074 | 0.2714 |
| 0.2652 | 5.0 | 1250 | 0.5017 | 0.0074 | 0.2281 |
| 0.1719 | 6.0 | 1500 | 0.4657 | 0.0074 | 0.2038 |
| 0.1011 | 7.0 | 1750 | 0.4561 | 0.0074 | 0.1905 |
| 0.0503 | 7.9685 | 1992 | 0.4572 | 0.0074 | 0.1710 |
Framework versions
- Transformers 4.48.3
- Pytorch 2.5.1+cu124
- Datasets 3.2.0
- Tokenizers 0.21.2