CoolFace
Modelpublic

Jubliano/wav2vec2-large-xls-r-300m-ipa

sourceHugging Faceupdated 2y agoView on Hugging Face
0likes22downloads
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-ipa

This model was trained from scratch on the commonvoice17_0 dataset. It achieves the following results on the evaluation set:

  • —Loss: 0.7309

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: 0.0001
  • —trainbatchsize: 6
  • —evalbatchsize: 8
  • —seed: 42
  • —gradientaccumulationsteps: 4
  • —totaltrainbatch_size: 24
  • —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • —lrschedulertype: linear
  • —lrschedulerwarmup_steps: 500
  • —num_epochs: 240
  • —mixedprecisiontraining: Native AMP

Training results

Training LossEpochStepValidation Loss
0.16773.68662001.0381
0.12147.37334000.5607
0.127211.05996000.5442
0.13514.74658000.5933
0.082418.433210000.6316
0.071122.119812000.5971
0.065325.806514000.6050
0.049929.493116000.6699
0.051633.179718000.6940
0.050736.866420000.7045
0.047840.553022000.7603
0.04544.239624000.7415
0.041947.926326000.7341
0.034451.612928000.7328
0.035455.299530000.8550
0.026858.986232000.7838
0.038362.672834000.7995
0.037166.359436000.7765
0.026470.046138000.8186
0.021273.732740000.7439
0.017777.419442000.7830
0.020481.106044000.8145
0.025484.792646000.8149
0.025788.479348000.7663
0.012692.165950000.7704
0.019695.852552000.7660
0.018599.539254000.8580
0.0236103.225856000.8169
0.0141106.912458000.8222
0.0142110.599160000.9001
0.0098114.285762000.8509
0.0372117.972464000.7734
0.0075121.659066000.8911
0.0118125.345668000.8347
0.0115129.032370000.8926
0.0164132.718972000.7985
0.006136.405574000.7571
0.0124140.092276000.8476
0.0141143.778878000.8071
0.0065147.465480000.7630
0.0095151.152182000.7161
0.0063154.838784000.8165
0.0107158.525386000.7411
0.0037162.212088000.7424
0.0045165.898690000.7611
0.0044169.585392000.7278
0.0043173.271994000.7396
0.0025176.958596000.7215
0.0029180.645298000.7551
0.0067184.3318100000.7518
0.0062188.0184102000.7668
0.0065191.7051104000.7433
0.0024195.3917106000.7942
0.0039199.0783108000.7448
0.0024202.7650110000.7290
0.0036206.4516112000.7678
0.0001210.1382114000.7390
0.0009213.8249116000.7292
0.0008217.5115118000.7383
0.0009221.1982120000.7435
0.0009224.8848122000.7324
0.0007228.5714124000.7444
0.0002232.2581126000.7228
0.0005235.9447128000.7309

Framework versions

  • —Transformers 4.40.2
  • —Pytorch 2.3.0+cu121
  • —Datasets 2.19.1
  • —Tokenizers 0.19.1
Jubliano/wav2vec2-large-xls-r-300m-ipa · CoolFace