metadata
language:
- ar
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- mozilla-foundation/common_voice_11_1
metrics:
- wer
model-index:
- name: Whisper Small Ar - Mhisham
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Common Voice 11.1 ar
type: mozilla-foundation/common_voice_11_1
config: ar
split: None
args: 'config: ar, split: test'
metrics:
- name: Wer
type: wer
value: 47.53466164723166
Whisper Small Ar - Mhisham
This model is a fine-tuned version of openai/whisper-small on the Common Voice 11.1 ar dataset. It achieves the following results on the evaluation set:
- Loss: 0.3242
- Wer: 47.5347
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: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 4000
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
0.2956 | 0.42 | 1000 | 0.3930 | 50.8659 |
0.2776 | 0.83 | 2000 | 0.3418 | 48.6604 |
0.1831 | 1.25 | 3000 | 0.3358 | 47.8175 |
0.1638 | 1.66 | 4000 | 0.3242 | 47.5347 |
Framework versions
- Transformers 4.39.2
- Pytorch 2.2.1+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2