metadata
language:
- multilingual
license: apache-2.0
base_model: serge-wilson/whisper-small-wolof
tags:
- generated_from_trainer
datasets:
- audiofolder
metrics:
- wer
model-index:
- name: Whisper Wolof Lengo AI V5
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: audiofolder
type: audiofolder
config: default
split: None
args: default
metrics:
- name: Wer
type: wer
value: 39.312847261594285
Whisper Wolof Lengo AI V5
This model is a fine-tuned version of serge-wilson/whisper-small-wolof on the audiofolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.3500
- Wer: 39.3128
- Cer: 26.3187
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: 0.0005
- train_batch_size: 8
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.98) and epsilon=1e-05
- lr_scheduler_type: polynomial
- lr_scheduler_warmup_steps: 50
- training_steps: 1990
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 1.3022 | 1.0 | 208 | 1.2112 | 146.8307 | 105.1707 |
| 0.8551 | 2.0 | 416 | 0.9020 | 90.5318 | 83.8833 |
| 0.5754 | 3.0 | 624 | 0.7127 | 118.9704 | 102.1064 |
| 0.3739 | 4.0 | 832 | 0.5951 | 63.1591 | 45.8295 |
| 0.2459 | 5.0 | 1040 | 0.4929 | 63.5446 | 50.1286 |
| 0.1579 | 6.0 | 1248 | 0.4524 | 51.6158 | 35.0170 |
| 0.0884 | 7.0 | 1456 | 0.4204 | 46.9554 | 30.6374 |
| 0.0498 | 8.0 | 1664 | 0.3817 | 51.6158 | 33.7194 |
| 0.0268 | 9.0 | 1872 | 0.3490 | 40.8550 | 27.1844 |
| 0.012 | 9.57 | 1990 | 0.3500 | 39.3128 | 26.3187 |
Framework versions
- Transformers 4.39.3
- Pytorch 2.2.2+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2