metadata
library_name: transformers
language:
- ar
license: apache-2.0
base_model: openai/whisper-base
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: Whisper base AR - BA
results: []
Whisper base AR - BA
This model is a fine-tuned version of openai/whisper-base on the quran-ayat-speech-to-text dataset. It achieves the following results on the evaluation set:
- Loss: 0.0034
- Wer: 0.0479
- Cer: 0.0195
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: 2
- total_train_batch_size: 32
- optimizer: Use OptimizerNames.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 |
|---|---|---|---|---|---|
| 0.0039 | 1.0 | 215 | 0.0034 | 0.0436 | 0.0167 |
| 0.0028 | 2.0 | 430 | 0.0039 | 0.0525 | 0.0204 |
| 0.0019 | 3.0 | 645 | 0.0051 | 0.0605 | 0.0231 |
| 0.0012 | 4.0 | 860 | 0.0054 | 0.0628 | 0.0232 |
| 0.0008 | 5.0 | 1075 | 0.0057 | 0.0648 | 0.0240 |
| 0.0006 | 6.0 | 1290 | 0.0061 | 0.0597 | 0.0212 |
| 0.0006 | 7.0 | 1505 | 0.0063 | 0.0621 | 0.0252 |
| 0.0004 | 8.0 | 1720 | 0.0073 | 0.0644 | 0.0251 |
| 0.0004 | 9.0 | 1935 | 0.0074 | 0.0621 | 0.0248 |
| 0.0002 | 10.0 | 2150 | 0.0081 | 0.0671 | 0.0253 |
| 0.0004 | 11.0 | 2365 | 0.0080 | 0.0632 | 0.0221 |
| 0.0002 | 12.0 | 2580 | 0.0083 | 0.0565 | 0.0207 |
| 0.0001 | 13.0 | 2795 | 0.0090 | 0.0570 | 0.0201 |
| 0.0001 | 14.0 | 3010 | 0.0105 | 0.0630 | 0.0263 |
| 0.0001 | 15.0 | 3225 | 0.0109 | 0.0608 | 0.0242 |
| 0.0001 | 16.0 | 3440 | 0.0118 | 0.0597 | 0.0221 |
| 0.0 | 17.0 | 3655 | 0.0119 | 0.0595 | 0.0220 |
| 0.0 | 18.0 | 3870 | 0.0130 | 0.0621 | 0.0235 |
| 0.0 | 19.0 | 4085 | 0.0133 | 0.0597 | 0.0231 |
| 0.0 | 20.0 | 4300 | 0.0133 | 0.0592 | 0.0240 |
| 0.0 | 21.0 | 4515 | 0.0135 | 0.0605 | 0.0237 |
| 0.0 | 22.0 | 4730 | 0.0135 | 0.0592 | 0.0231 |
| 0.0 | 23.0 | 4945 | 0.0135 | 0.0592 | 0.0231 |
| 0.0 | 24.0 | 5160 | 0.0124 | 0.0589 | 0.0228 |
| 0.0 | 25.0 | 5375 | 0.0135 | 0.0590 | 0.0231 |
Framework versions
- Transformers 4.51.1
- Pytorch 2.5.1+cu124
- Datasets 3.5.0
- Tokenizers 0.21.0