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: 0.0121
- Wer: 13.2495
- Cer: 4.3278
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: 0.0001
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 4
- 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: cosine
- lr_scheduler_warmup_steps: 500
- num_epochs: 2
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Cer | Validation Loss | Wer |
|---|---|---|---|---|---|
| 0.0258 | 0.1408 | 400 | 52.2218 | 0.0246 | 104.9348 |
| 0.0177 | 0.2817 | 800 | 10.2633 | 0.0184 | 26.2089 |
| 0.0116 | 0.4225 | 1200 | 7.3210 | 0.0160 | 20.9517 |
| 0.0101 | 0.5633 | 1600 | 5.8227 | 0.0141 | 17.5020 |
| 0.008 | 0.7042 | 2000 | 5.1235 | 0.0127 | 16.3695 |
| 0.0057 | 0.8450 | 2400 | 4.8168 | 0.0119 | 15.2343 |
| 0.0056 | 0.9858 | 2800 | 4.6678 | 0.0116 | 14.6364 |
| 0.0071 | 1.1267 | 3200 | 0.0135 | 15.8929 | 5.3042 |
| 0.0059 | 1.2676 | 3600 | 0.0132 | 15.7165 | 5.0437 |
| 0.0056 | 1.4084 | 4000 | 0.0124 | 14.5758 | 5.3648 |
| 0.0041 | 1.5492 | 4400 | 0.0122 | 14.2259 | 4.7531 |
| 0.0038 | 1.6901 | 4800 | 0.0120 | 13.8043 | 4.7431 |
| 0.004 | 1.8309 | 5200 | 0.0119 | 14.1818 | 4.9569 |
| 0.0036 | 1.9717 | 5600 | 0.0118 | 14.0743 | 4.9171 |
Framework versions
- Transformers 4.47.0
- Pytorch 2.5.1+cu121
- Datasets 3.2.0
- Tokenizers 0.21.0