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: 0.1055
- Wer: 0.2245
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: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- 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: 15
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 4.1388 | 1.0 | 313 | 0.1088 | 0.2255 |
| 4.005 | 2.0 | 626 | 0.1127 | 0.2380 |
| 3.1681 | 3.0 | 939 | 0.1117 | 0.2215 |
| 2.4917 | 4.0 | 1252 | 0.1089 | 0.2202 |
| 2.1826 | 5.0 | 1565 | 0.1062 | 0.2146 |
| 1.9244 | 6.0 | 1878 | 0.1062 | 0.2263 |
| 1.7281 | 7.0 | 2191 | 0.1032 | 0.2188 |
| 1.5604 | 8.0 | 2504 | 0.1032 | 0.2193 |
| 1.5071 | 9.0 | 2817 | 0.1031 | 0.2244 |
| 1.3603 | 10.0 | 3130 | 0.1033 | 0.2122 |
| 1.2858 | 11.0 | 3443 | 0.1022 | 0.2134 |
| 1.1788 | 12.0 | 3756 | 0.1024 | 0.2106 |
| 1.1271 | 13.0 | 4069 | 0.1015 | 0.2121 |
| 1.0559 | 14.0 | 4382 | 0.1017 | 0.2098 |
| 1.0875 | 14.9536 | 4680 | 0.1013 | 0.2103 |
Framework versions
- Transformers 4.51.1
- Pytorch 2.5.1+cu124
- Datasets 3.5.0
- Tokenizers 0.21.0