metadata
language:
- ara
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- google/fleurs
metrics:
- wer
model-index:
- name: Whisper Small Ar_Eg
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Fleurs ar_eg
type: google/fleurs
config: ar_eg
split: None
args: 'config: ara, split: test'
metrics:
- name: Wer
type: wer
value: 23.1
Whisper Small Ar_Eg
This model is a fine-tuned version of openai/whisper-small on the Fleurs ar_eg dataset. It achieves the following results on the evaluation set:
- Loss: 0.4820
- Wer: 23.1
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 |
|---|---|---|---|---|
| 0.058 | 6.6667 | 1000 | 0.3934 | 23.6625 |
| 0.0014 | 13.3333 | 2000 | 0.4452 | 22.9875 |
| 0.0005 | 20.0 | 3000 | 0.4719 | 22.9375 |
| 0.0004 | 26.6667 | 4000 | 0.4820 | 23.1 |
Framework versions
- Transformers 4.40.2
- Pytorch 2.1.2
- Datasets 2.19.1
- Tokenizers 0.19.1