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: 0.0097
- Wer: 0.0951
- Cer: 0.0412
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-06
- 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: 20
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 0.0078 | 1.0 | 313 | 0.0073 | 0.0916 | 0.0336 |
| 0.0059 | 2.0 | 626 | 0.0071 | 0.0878 | 0.0325 |
| 0.0051 | 3.0 | 939 | 0.0072 | 0.0878 | 0.0331 |
| 0.0036 | 4.0 | 1252 | 0.0077 | 0.0874 | 0.0336 |
| 0.0017 | 5.0 | 1565 | 0.0081 | 0.0881 | 0.0341 |
| 0.0018 | 6.0 | 1878 | 0.0086 | 0.0871 | 0.0357 |
| 0.0016 | 7.0 | 2191 | 0.0091 | 0.0909 | 0.0369 |
| 0.0016 | 8.0 | 2504 | 0.0095 | 0.0842 | 0.0327 |
| 0.0011 | 9.0 | 2817 | 0.0098 | 0.0889 | 0.0336 |
| 0.001 | 10.0 | 3130 | 0.0102 | 0.0872 | 0.0415 |
| 0.0008 | 11.0 | 3443 | 0.0104 | 0.0865 | 0.0321 |
Framework versions
- Transformers 4.47.0
- Pytorch 2.5.1+cu121
- Datasets 3.2.0
- Tokenizers 0.21.0