metadata
library_name: transformers
language:
- ar
license: apache-2.0
base_model: openai/whisper-tiny
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: Whisper tiny AR - BH
results: []
Whisper tiny AR - BH
This model is a fine-tuned version of openai/whisper-tiny on the quran-ayat-speech-to-text dataset. It achieves the following results on the evaluation set:
- Loss: 0.0278
- Wer: 0.8483
- Cer: 0.4043
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: 5e-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: 500
- num_epochs: 25
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 0.0691 | 1.0 | 157 | 0.0645 | 4.6639 | 1.8253 |
| 0.0429 | 2.0 | 314 | 0.0408 | 1.8489 | 0.8577 |
| 0.0287 | 3.0 | 471 | 0.0332 | 2.3785 | 1.2099 |
| 0.0224 | 4.0 | 628 | 0.0293 | 1.3376 | 0.7609 |
| 0.012 | 5.0 | 785 | 0.0276 | 1.8050 | 0.9207 |
| 0.0098 | 6.0 | 942 | 0.0273 | 2.0293 | 1.0277 |
| 0.0076 | 7.0 | 1099 | 0.0272 | 1.1354 | 0.5388 |
| 0.0068 | 8.0 | 1256 | 0.0288 | 0.6950 | 0.3135 |
| 0.0042 | 9.0 | 1413 | 0.0292 | 0.9439 | 0.4304 |
| 0.003 | 10.0 | 1570 | 0.0290 | 1.5183 | 0.7840 |
| 0.0024 | 11.0 | 1727 | 0.0301 | 0.8514 | 0.3949 |
Framework versions
- Transformers 4.47.0
- Pytorch 2.5.1+cu121
- Datasets 3.2.0
- Tokenizers 0.21.0