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: 0.0171
- Wer: 0.1132
- Cer: 0.0409
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: Adam with betas=(0.9,0.999) and epsilon=1e-08
- 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 |
|---|---|---|---|---|---|
| 0.0145 | 1.0 | 219 | 0.0160 | 0.1213 | 0.0455 |
| 0.0115 | 2.0 | 438 | 0.0123 | 0.1267 | 0.0469 |
| 0.0082 | 3.0 | 657 | 0.0112 | 0.1157 | 0.0399 |
| 0.0061 | 4.0 | 876 | 0.0111 | 0.1173 | 0.0406 |
| 0.0056 | 5.0 | 1095 | 0.0116 | 0.1137 | 0.0389 |
| 0.0062 | 6.0 | 1314 | 0.0122 | 0.1117 | 0.0367 |
| 0.0021 | 7.0 | 1533 | 0.0129 | 0.1166 | 0.0396 |
| 0.0028 | 8.0 | 1752 | 0.0136 | 0.1167 | 0.0385 |
| 0.0017 | 9.0 | 1971 | 0.0141 | 0.1140 | 0.0370 |
| 0.0017 | 10.0 | 2190 | 0.0148 | 0.1119 | 0.0381 |
| 0.0012 | 11.0 | 2409 | 0.0152 | 0.1117 | 0.0366 |
| 0.0013 | 12.0 | 2628 | 0.0156 | 0.1122 | 0.0373 |
| 0.0008 | 13.0 | 2847 | 0.0159 | 0.1120 | 0.0377 |
| 0.0011 | 14.0 | 3066 | 0.0171 | 0.1128 | 0.0408 |
| 0.0008 | 15.0 | 3285 | 0.0161 | 0.1108 | 0.0369 |
Framework versions
- Transformers 4.44.2
- Pytorch 2.4.1+cu121
- Datasets 3.2.0
- Tokenizers 0.19.1