pulaar_stt_1
This model is a fine-tuned version of openai/whisper-small on the abdouaziz/pulaar_concat_V3 dataset. It achieves the following results on the evaluation set:
- Loss: 0.5820
- Wer: 0.3903
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: 3e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 2
- 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: 50
- training_steps: 48000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.9469 | 3.3557 | 2000 | 0.5820 | 0.3903 |
| 0.1404 | 6.7114 | 4000 | 0.6877 | 0.3340 |
| 0.047 | 10.0671 | 6000 | 0.7400 | 0.3206 |
| 0.0298 | 13.4228 | 8000 | 0.7731 | 0.3272 |
| 0.0231 | 16.7785 | 10000 | 0.8005 | 0.3239 |
Framework versions
- Transformers 4.46.0
- Pytorch 2.9.0+cu128
- Datasets 4.3.0
- Tokenizers 0.20.3