whisper-tiny-ar / README.md
saralameri's picture
End of training
cca0d1b verified
|
raw
history blame
2.13 kB
metadata
library_name: transformers
language:
  - ar
license: apache-2.0
base_model: openai/whisper-tiny
tags:
  - generated_from_trainer
datasets:
  - mozilla-foundation/common_voice_16_1
metrics:
  - wer
model-index:
  - name: Whisper Tiny Ar
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Common Voice 16.1
          type: mozilla-foundation/common_voice_16_1
          config: ar
          split: test
          args: 'config: ar, split: test'
        metrics:
          - name: Wer
            type: wer
            value: 67.60586678429644

Whisper Tiny Ar

This model is a fine-tuned version of openai/whisper-tiny on the Common Voice 16.1 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.6267
  • Wer: 67.6059

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: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • 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
0.6119 0.4122 1000 0.7503 75.1415
0.5206 0.8244 2000 0.6736 70.3904
0.4317 1.2366 3000 0.6404 67.5599
0.411 1.6488 4000 0.6267 67.6059

Framework versions

  • Transformers 4.48.1
  • Pytorch 2.5.1+cu124
  • Datasets 3.2.0
  • Tokenizers 0.21.0