whisper-small-th / README.md
napatswift's picture
Update README.md
77beda0
metadata
language:
  - th
license: apache-2.0
tags:
  - whisper-event
  - generated_from_trainer
datasets:
  - mozilla-foundation/common_voice_11_0
metrics:
  - wer
model-index:
  - name: Whisper Small Thai
    results:
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: mozilla-foundation/common_voice_11_0 th
          type: mozilla-foundation/common_voice_11_0
          config: th
          split: test
          args: th
        metrics:
          - type: wer
            value: 14.060702592690912
            name: Wer
          - type: mer
            value: 13.786820528393562
            name: Mer

Whisper Small Thai

This model is a fine-tuned version of openai/whisper-small on the mozilla-foundation/common_voice_11_0 th,None,th_th dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1841
  • Wer: 14.060

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: 0.0001
  • train_batch_size: 64
  • eval_batch_size: 32
  • 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
0.0909 0.2 1000 0.3373 25.5752
0.0426 1.1 2000 0.2540 20.9739
0.0267 2.0 3000 0.2210 17.4080
0.0145 2.2 4000 0.2134 15.5675
0.0099 3.1 5000 0.1841 13.2285

Framework versions

  • Transformers 4.26.0.dev0
  • Pytorch 1.13.1+cu117
  • Datasets 2.7.1.dev0
  • Tokenizers 0.13.2