whisper small finetuned speed augmentation TLT non-native child speech
This model is a fine-tuned version of openai/whisper-small on the LTL2021 dataset. It achieves the following results on the evaluation set:
- Loss: 0.4043
- Wer: 19.4862
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
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- 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: 2048
- training_steps: 4000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 5.2971 | 1.6087 | 500 | 2.9974 | 17.8041 |
| 3.2025 | 3.2158 | 1000 | 1.8702 | 17.0941 |
| 1.7907 | 4.8245 | 1500 | 1.0121 | 18.1471 |
| 0.8310 | 6.4316 | 2000 | 0.4578 | 18.5425 |
| 0.6474 | 8.0386 | 2500 | 0.4194 | 18.9138 |
| 0.6284 | 9.6473 | 3000 | 0.4095 | 19.6063 |
| 0.6056 | 11.2544 | 3500 | 0.4054 | 20.4168 |
| 0.5855 | 12.8631 | 4000 | 0.4043 | 19.4862 |
Framework versions
- Transformers 5.8.1
- Pytorch 2.5.1+cu121
- Datasets 4.8.5
- Tokenizers 0.22.2