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: 0.0094
- Wer: 0.0890
- Cer: 0.0345
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: 15
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 0.007 | 1.0 | 469 | 0.0077 | 0.0912 | 0.0333 |
| 0.0066 | 2.0 | 938 | 0.0070 | 0.0890 | 0.0329 |
| 0.0039 | 3.0 | 1407 | 0.0071 | 0.0867 | 0.0320 |
| 0.0038 | 4.0 | 1876 | 0.0075 | 0.0872 | 0.0332 |
| 0.0029 | 5.0 | 2345 | 0.0080 | 0.0854 | 0.0314 |
| 0.0013 | 6.0 | 2814 | 0.0086 | 0.0840 | 0.0317 |
| 0.002 | 7.0 | 3283 | 0.0089 | 0.0858 | 0.0313 |
| 0.0008 | 8.0 | 3752 | 0.0094 | 0.0843 | 0.0321 |
| 0.0012 | 9.0 | 4221 | 0.0099 | 0.0852 | 0.0318 |
| 0.0008 | 10.0 | 4690 | 0.0103 | 0.0854 | 0.0321 |
| 0.0009 | 11.0 | 5159 | 0.0107 | 0.0871 | 0.0325 |
| 0.0006 | 12.0 | 5628 | 0.0109 | 0.0860 | 0.0325 |
| 0.0007 | 13.0 | 6097 | 0.0111 | 0.1678 | 0.0709 |
| 0.0006 | 14.0 | 6566 | 0.0125 | 0.1149 | 0.0466 |
| 0.0004 | 15.0 | 7035 | 0.0113 | 0.1645 | 0.0699 |
Framework versions
- Transformers 4.47.0
- Pytorch 2.5.1+cu121
- Datasets 3.2.0
- Tokenizers 0.21.0