Whisper large LoRA Merged Es - Jbautistas
This model is a fine-tuned version of openai/whisper-large-v3 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.7153
- Wer: 58.9379
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: 0.0001
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 32
- optimizer: Use 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: 500
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 1.4943 | 6.2759 | 50 | 1.1849 | 58.2648 |
| 1.1692 | 12.5517 | 100 | 1.0537 | 56.2453 |
| 0.9958 | 18.8276 | 150 | 0.8947 | 54.8990 |
| 0.7308 | 25.0 | 200 | 0.7153 | 58.9379 |
Framework versions
- Transformers 4.56.0
- Pytorch 2.3.0+cu121
- Datasets 3.1.0
- Tokenizers 0.22.0