metadata
library_name: transformers
language:
- ar
license: apache-2.0
base_model: openai/whisper-small
tags:
- automatic-speech-recognition
- arabic
- generated_from_trainer
datasets:
- audiofolder
metrics:
- wer
model-index:
- name: Whisper Small Informal Arabic
results:
- task:
type: automatic-speech-recognition
name: Automatic Speech Recognition
dataset:
name: Informal Arabic
type: audiofolder
config: default
split: None
args: default
metrics:
- type: wer
value: 35.868522072936656
name: Wer
Whisper Small Informal Arabic
This model is a fine-tuned version of openai/whisper-small on the Informal Arabic dataset. It achieves the following results on the evaluation set:
- Loss: 0.7928
- Wer: 35.8685
- Cer: 12.0407
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: 1e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- 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
- training_steps: 5000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 0.0019 | 26.3158 | 1000 | 0.6358 | 37.4280 | 12.6300 |
| 0.0004 | 52.6316 | 2000 | 0.6972 | 35.4726 | 12.0947 |
| 0.0002 | 78.9474 | 3000 | 0.7480 | 35.8925 | 12.1017 |
| 0.0001 | 105.2632 | 4000 | 0.7751 | 35.9165 | 12.0853 |
| 0.0001 | 131.5789 | 5000 | 0.7928 | 35.8685 | 12.0407 |
Framework versions
- Transformers 4.50.3
- Pytorch 2.6.0+cu124
- Datasets 3.5.0
- Tokenizers 0.21.1