CoolFace
Modelpublic

yesj1234/xlsr_mid2_ko-en

sourceHugging Faceapache-2.0updated 3y agoView on Hugging Face
0likes5downloads
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. -->

en-xlsr

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the ./SAMPLE_SPEECH.PY - NA dataset. It achieves the following results on the evaluation set:

  • —Loss: 0.3889
  • —Cer: 0.1082

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: 4
  • —evalbatchsize: 4
  • —seed: 42
  • —distributed_type: multi-GPU
  • —num_devices: 4
  • —gradientaccumulationsteps: 4
  • —totaltrainbatch_size: 64
  • —totalevalbatch_size: 16
  • —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • —lrschedulertype: linear
  • —lrschedulerwarmup_steps: 1500
  • —num_epochs: 15

Training results

Training LossEpochStepValidation LossCer
1.45031.2220001.06100.2687
1.02392.4540000.69620.1904
0.89773.6760000.59450.1687
0.8044.980000.53280.1492
0.6986.12100000.50140.1365
0.64267.35120000.47150.1322
0.618.57140000.45300.1258
0.57099.79160000.43000.1201
0.523511.02180000.41680.1166
0.477812.24200000.40570.1129
0.457113.47220000.39450.1100
0.438814.69240000.38910.1081

Framework versions

  • —Transformers 4.34.0
  • —Pytorch 2.1.0+cu121
  • —Datasets 2.14.5
  • —Tokenizers 0.14.1