metadata
language:
- vi
license: apache-2.0
library_name: peft
tags:
- generated_from_trainer
base_model: openai/whisper-small
datasets:
- stevehoang9/vietnamese_ASR
metrics:
- wer
model-index:
- name: Whisper Small Vietnam - Steve Hoang
results:
- task:
type: automatic-speech-recognition
name: Automatic Speech Recognition
dataset:
name: VIVOS
type: stevehoang9/vietnamese_ASR
metrics:
- type: wer
value: 51.68350168350169
name: Wer
Whisper Small Vietnam - Steve Hoang
This model is a fine-tuned version of openai/whisper-small on the VIVOS dataset. It achieves the following results on the evaluation set:
- Loss: 0.6333
- Wer: 51.6835
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: 5000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.931 | 1.3717 | 1000 | 0.9599 | 74.7216 |
| 0.7286 | 2.7435 | 2000 | 0.7709 | 61.5255 |
| 0.6341 | 4.1152 | 3000 | 0.6854 | 55.5685 |
| 0.6044 | 5.4870 | 4000 | 0.6458 | 52.8490 |
| 0.6031 | 6.8587 | 5000 | 0.6333 | 51.6835 |
Framework versions
- PEFT 0.11.1
- Transformers 4.41.2
- Pytorch 2.1.2
- Datasets 2.19.2
- Tokenizers 0.19.1