CoolFace
Modelpublic

jonatasgrosman/whisper-small-pt-cv11-v6

sourceHugging Faceapache-2.0updated 4y agoView on Hugging Face
0likes12downloads
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 Portuguese

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

  • —Loss: 0.2738
  • —Wer: 11.9723
  • —Cer: 4.8273

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: 5e-06
  • —trainbatchsize: 32
  • —evalbatchsize: 16
  • —seed: 42
  • —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • —lrschedulertype: linear
  • —lrschedulerwarmup_steps: 1000
  • —training_steps: 10000
  • —mixedprecisiontraining: Native AMP

Training results

Training LossEpochStepValidation LossWerCer
0.52350.925000.360515.40836.1205
0.38391.8410000.303414.28355.6010
0.28282.7615000.285213.49775.1727
0.23673.6820000.276812.91225.2280
0.18324.625000.272812.24964.9157
0.15495.5230000.273012.06474.8384
0.13186.4535000.275712.09554.8135
0.10777.3740000.273811.97234.8273
0.09698.2945000.278412.15724.9212
0.08139.2150000.280512.31125.0207
0.075110.1355000.283112.08014.8494

Framework versions

  • —Transformers 4.26.0.dev0
  • —Pytorch 1.12.1+cu116
  • —Datasets 2.7.1.dev0
  • —Tokenizers 0.13.2