whisper-small-ar / README.md
MMars's picture
Update README.md
28e435a
metadata
license: apache-2.0
tags:
  - whisper-event
  - generated_from_trainer
  - hf-asr-leaderboard
datasets:
  - mozilla-foundation/common_voice_11_0
metrics:
  - wer
model-index:
  - name: Whisper-small-ar - Mourad Mars
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: mozilla-foundation/common_voice_11_0
          type: mozilla-foundation/common_voice_11_0
          config: ar
          split: test
          args: ar
        metrics:
          - name: Wer
            type: wer
            value: 44.976586

Whisper-small-ar - Mourad Mars

This model is a fine-tuned version of openai/whisper-small on the common_voice_11_0 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.322550
  • Wer: 44.976586

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:

train_batch_size=16

eval_batch_size=8

optimizer: Adam

learning_rate=1e-5

warmup_steps=500

max_steps=4000
    
eval_steps=1000


metric_for_best_model="wer"

Training results

Training Loss Step Validation Loss Wer
0.2811 1000 0.393018 53.778349
0.2356 2000 0.348794 47.793591
0.1705 3000 0.332207 45.758883
0.1476 4000 0.322550 44.976586

Framework versions