CoolFace
Modelpublic

whispy/whisper_italian

sourceHugging Faceapache-2.0updated 4y agoView on Hugging Face
9likes18downloads
Model Card

Whisper Small Italian

This model is a fine-tuned version of openai/whisper-base on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:

  • —Loss: 0.1185
  • —Wer: 17.3916

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: 16
  • —evalbatchsize: 8 <!-- - seed: 42 -->
  • —gradientaccumulationsteps: 1
  • —totaltrainbatch_size: 16
  • —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • —lrschedulertype: linear
  • —lrschedulerwarmup_steps: 500
  • —trainingsteps: 954 <!-- - mixedprecision_training: Native AMP -->

Training results

Training LossStepValidation LossWer
1.47441001.1852117.6059
0.72412000.745279.7386
0.33213000.321521.0497
0.29304000.303020.2129
0.26985000.298219.7635
0.24536000.289819.0097
0.23387000.276818.7054
0.24028000.264618.2214
0.23409000.258117.3916

Framework versions

  • —Transformers 4.25.0.dev0
  • —Pytorch 1.12.1+cu113
  • —Datasets 2.7.1
  • —Tokenizers 0.13.2