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: 0.0986
- Wer: 0.2011
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 |
|---|---|---|---|---|
| 2.5906 | 1.0 | 313 | 0.0946 | 0.1988 |
| 2.3807 | 2.0 | 626 | 0.0971 | 0.2019 |
| 1.8335 | 3.0 | 939 | 0.0947 | 0.1939 |
| 1.6071 | 4.0 | 1252 | 0.0959 | 0.1995 |
| 1.4632 | 5.0 | 1565 | 0.0957 | 0.2039 |
| 1.3703 | 6.0 | 1878 | 0.0940 | 0.2041 |
| 1.2553 | 7.0 | 2191 | 0.0947 | 0.1961 |
| 1.0844 | 8.0 | 2504 | 0.0939 | 0.1973 |
| 1.0211 | 9.0 | 2817 | 0.0929 | 0.1997 |
| 0.959 | 10.0 | 3130 | 0.0931 | 0.1989 |
| 0.9041 | 11.0 | 3443 | 0.0925 | 0.1986 |
| 0.8402 | 12.0 | 3756 | 0.0924 | 0.1983 |
| 0.8244 | 13.0 | 4069 | 0.0917 | 0.1989 |
| 0.7911 | 14.0 | 4382 | 0.0917 | 0.2000 |
| 0.7672 | 14.9536 | 4680 | 0.0917 | 0.1997 |
Framework versions
- Transformers 4.51.1
- Pytorch 2.5.1+cu124
- Datasets 3.5.0
- Tokenizers 0.21.0