whisper-small-tamil / README.md
carlfeynman's picture
Update README.md
e9bc5f9 verified
|
raw
history blame
2.09 kB
metadata
language:
  - ta
license: apache-2.0
widgets:
  - label: Example 1
    audio: https://yourdomain.com/path/to/example1.wav
  - label: Example 2
    audio: https://yourdomain.com/path/to/example2.wav
tags:
  - generated_from_trainer
datasets:
  - mozilla-foundation/common_voice_13_0
metrics:
  - wer
base_model: openai/whisper-small
model-index:
  - name: carl-whisper-small-finetuned-tamil
    results:
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: Common Voice 13
          type: mozilla-foundation/common_voice_13_0
          config: ta
          split: None
          args: ta
        metrics:
          - type: wer
            value: 21.830330026321118
            name: Wer

carl-whisper-small-finetuned-tamil

This model is a fine-tuned version of openai/whisper-small on the Common Voice 13 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.4660
  • Wer Ortho: 62.9625
  • Wer: 21.8303

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: 16
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: constant_with_warmup
  • lr_scheduler_warmup_steps: 10
  • training_steps: 100
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer Ortho Wer
0.3504 0.37 100 0.4660 62.9625 21.8303

Framework versions

  • Transformers 4.38.2
  • Pytorch 1.11.0+cu102
  • Datasets 2.18.0
  • Tokenizers 0.15.2