whisper-base-SER-v5_1
This fine-tune is corrupted (because i used mistakenly only 100 rows for training😑😑😑
)
This model is a fine-tuned version of openai/whisper-base on the Whisper_Compatible_SER_benchmark(Not train_augmented) dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3675
- Wer: 236.0
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: 8
- seed: 42
- 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
- training_steps: 6000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0004 | 250.0 | 1000 | 0.2744 | 665.0 |
| 0.0001 | 500.0 | 2000 | 0.3142 | 413.0 |
| 0.0 | 750.0 | 3000 | 0.3356 | 239.0 |
| 0.0 | 1000.0 | 4000 | 0.3451 | 239.0 |
| 0.0 | 1250.0 | 5000 | 0.3657 | 236.0 |
| 0.0 | 1500.0 | 6000 | 0.3675 | 236.0 |
Framework versions
- Transformers 4.48.0
- Pytorch 2.5.1+cu121
- Datasets 3.2.0
- Tokenizers 0.21.0
- Downloads last month
- 2
Model tree for iFaz/whisper-base-SER-v5_2
Base model
openai/whisper-baseDataset used to train iFaz/whisper-base-SER-v5_2
Evaluation results
- Wer on Whisper_Compatible_SER_benchmark(Not train_augmented)self-reported236.000