CoolFace
Modelpublic

nrshoudi/wav2vec2-large-xls-r-300m-Arabic-MDD

sourceHugging Faceapache-2.0updated 3y agoView on Hugging Face
0likes40downloads
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-Arabic-MDD

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the None dataset. It achieves the following results on the evaluation set:

  • —Loss: 0.0204
  • —Per: 0.0096

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: 0.0005
  • —trainbatchsize: 16
  • —evalbatchsize: 6
  • —seed: 42
  • —gradientaccumulationsteps: 4
  • —totaltrainbatch_size: 64
  • —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • —lrschedulertype: linear
  • —lrschedulerwarmup_steps: 250
  • —num_epochs: 20.0

Training results

Training LossEpochStepValidation LossPer
8.48921.0513.32871.0
2.56061.991022.04721.0
2.05232.991532.01841.0
1.98654.02051.98651.0
1.98545.02561.94080.9934
1.96035.993071.91240.9937
1.81696.993581.69650.9871
1.55968.04101.48500.9779
1.34949.04611.10740.9782
0.92429.995120.60710.7359
0.517210.995630.29100.2875
0.310512.06150.14890.1075
0.18613.06660.09380.0342
0.143713.997170.08090.0255
0.118514.997680.06650.0127
0.095216.08200.04090.0103
0.072117.08710.02910.0107
0.057517.999220.02280.0089
0.050318.999730.02160.0100
0.043919.910200.02040.0096

Framework versions

  • —Transformers 4.30.1
  • —Pytorch 2.0.1+cu118
  • —Datasets 1.18.3
  • —Tokenizers 0.13.3