CoolFace
Modelpublic

bilalahmed15/Urdu_repo

sourceHugging Faceapache-2.0updated 4y agoView on Hugging Face
0likes64downloads
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. -->

wav2vec2-large-xls-r-300m-turkish-colab

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice dataset. It achieves the following results on the evaluation set:

  • —Loss: 0.7532
  • —Wer: 0.4020

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: 0.0003
  • —trainbatchsize: 16
  • —evalbatchsize: 8
  • —seed: 42
  • —gradientaccumulationsteps: 2
  • —totaltrainbatch_size: 32
  • —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • —lrschedulertype: linear
  • —lrschedulerwarmup_steps: 500
  • —num_epochs: 30
  • —mixedprecisiontraining: Native AMP

Training results

Training LossEpochStepValidation LossWer
5.95421.964001.57370.8827
0.85963.928000.72960.5696
0.47295.8812000.60040.4934
0.33647.8416000.57760.4656
0.26849.820000.61780.4563
0.214311.7624000.64080.4690
0.174413.7228000.67040.4573
0.145815.6832000.70150.4484
0.120117.6536000.71510.4228
0.10419.6140000.71230.4195
0.088721.5744000.71020.4234
0.080723.5348000.75610.4132
0.069725.4952000.74350.4075
0.061127.4556000.74650.4034
0.055629.4160000.75320.4020

Framework versions

  • —Transformers 4.18.0
  • —Pytorch 1.11.0
  • —Datasets 2.1.0
  • —Tokenizers 0.12.1