README.md

metadata

base_model: openai/whisper-small-v3
datasets:
  - mozilla-foundation/common_voice_11_0
language:
  - vi
library_name: transformers
metrics:
  - wer
tags:
  - generated_from_trainer
model-index:
  - name: Whisper small vi - Ox
    results:
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: Common Voice 11.0
          type: mozilla-foundation/common_voice_11_0
          config: vi
          split: test
          args: 'config: vi, split: test'
        metrics:
          - type: wer
            value: 14.738458164272398
            name: Wer

Whisper small vi - Ox

This model is a fine-tuned version of openai/whisper-small-v3 on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.2529
• Wer: 14.7385

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
0.2196	1.3928	1000	0.3174	19.4758
0.0938	2.7855	2000	0.2513	16.0325
0.014	4.1783	3000	0.2467	14.4972
0.0109	5.5710	4000	0.2529	14.7385

Framework versions

• Transformers 4.45.2
• Pytorch 2.4.1
• Datasets 3.0.2
• Tokenizers 0.20.1