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: 0.1029
- Wer: 0.2133
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 |
|---|---|---|---|---|
| 2.6022 | 1.0 | 154 | 0.0981 | 0.2051 |
| 2.386 | 2.0 | 308 | 0.0994 | 0.2067 |
| 1.9412 | 3.0 | 462 | 0.0996 | 0.2022 |
| 1.824 | 4.0 | 616 | 0.0998 | 0.2045 |
| 1.6295 | 5.0 | 770 | 0.0976 | 0.2101 |
| 1.458 | 6.0 | 924 | 0.0982 | 0.2007 |
| 1.2992 | 7.0 | 1078 | 0.0959 | 0.2063 |
| 1.1839 | 8.0 | 1232 | 0.0971 | 0.2088 |
| 1.0944 | 9.0 | 1386 | 0.0954 | 0.2085 |
| 1.032 | 10.0 | 1540 | 0.0956 | 0.2104 |
| 0.9384 | 11.0 | 1694 | 0.0948 | 0.2026 |
| 0.8799 | 12.0 | 1848 | 0.0942 | 0.2053 |
| 0.8374 | 13.0 | 2002 | 0.0939 | 0.2062 |
| 0.7891 | 14.0 | 2156 | 0.0940 | 0.2065 |
| 0.7622 | 14.9070 | 2295 | 0.0940 | 0.2050 |
Framework versions
- Transformers 4.51.1
- Pytorch 2.5.1+cu124
- Datasets 3.5.0
- Tokenizers 0.21.0