CoolFace
Modelpublic

lorenzoncina/whisper-small-ar

sourceHugging Faceapache-2.0updated 3y agoView on Hugging Face
0likes15downloads
Model Card

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

Whisper Small Arabic

This model is a fine-tuned version of openai/whisper-small on the mozilla-foundation/commonvoice11_0 ar dataset. It achieves the following results on the evaluation set:

  • —Loss: 0.4948
  • —Wer: 54.08

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
  • —trainbatchsize: 32
  • —evalbatchsize: 2
  • —seed: 42
  • —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • —lrschedulertype: linear
  • —lrschedulerwarmup_steps: 500
  • —training_steps: 10000
  • —mixedprecisiontraining: Native AMP

Training results

Training LossEpochStepValidation LossWer
0.18851.0310000.395066.44
0.07943.020000.395058.5507
0.02864.0430000.460263.88
0.01286.0140000.494854.08
0.00487.0450000.546657.9867
0.00299.0160000.571055.4147
0.001310.0570000.599658.7707
0.000812.0280000.617954.748
0.000613.0590000.634356.2613
0.000315.02100000.638856.228

Framework versions

  • —Transformers 4.28.0.dev0
  • —Pytorch 2.0.0+cu117
  • —Datasets 2.11.1.dev0
  • —Tokenizers 0.13.2