Whisper Small Informal Arabic

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

  • Loss: 0.7988
  • Wer: 35.7244
  • Cer: 13.7759

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 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: 1000
  • training_steps: 5000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer Cer
0.1677 4.6296 1000 0.5868 38.8724 15.6307
0.0075 9.2593 2000 0.6943 37.7629 15.0868
0.0009 13.8889 3000 0.7523 35.4793 13.7734
0.0007 18.5185 4000 0.7892 35.5180 13.7529
0.0003 23.1481 5000 0.7988 35.7244 13.7759

Framework versions

  • Transformers 4.49.0
  • Pytorch 2.5.1+cu118
  • Datasets 3.3.2
  • Tokenizers 0.21.0
Downloads last month
6
Safetensors
Model size
242M params
Tensor type
F32
·
Inference Providers NEW
This model is not currently available via any of the supported Inference Providers.

Model tree for Talyiamira/whisper-small-informal-arabic-new

Finetuned
(2368)
this model

Evaluation results