CoolFace
Modelpublic

utakumi/Hubert-common_voice-ja-demo-kana-only-cosine

sourceHugging Faceapache-2.0updated 2y agoView on Hugging Face
0likes10downloads
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. -->

Hubert-common_voice-ja-demo-kana-only-cosine

This model is a fine-tuned version of rinna/japanese-hubert-base on the MOZILLA-FOUNDATION/COMMONVOICE13_0 - JA dataset. It achieves the following results on the evaluation set:

  • —Loss: 0.6322
  • —Wer: 1.0
  • —Cer: 0.3308

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: 3e-05
  • —trainbatchsize: 16
  • —evalbatchsize: 8
  • —seed: 42
  • —gradientaccumulationsteps: 2
  • —totaltrainbatch_size: 32
  • —optimizer: Use adamwtorch with betas=(0.9,0.999) and epsilon=1e-08 and optimizerargs=No additional optimizer arguments
  • —lrschedulertype: cosine
  • —lrschedulerwarmup_steps: 12500
  • —num_epochs: 25.0
  • —mixedprecisiontraining: Native AMP

Training results

Training LossEpochStepValidation LossWerCer
No log0.266010043.22511.52955.8210
No log0.531920042.44521.53225.2849
No log0.797930040.41541.12621.8126
No log1.063840032.80211.00.9999
31.8841.329850020.81331.00.9999
31.8841.595760017.58241.00.9999
31.8841.861770016.86821.00.9999
31.8842.127780016.44661.00.9999
31.8842.393690016.00371.00.9999
14.47012.6596100015.54091.00.9999
14.47012.9255110015.04461.00.9999
14.47013.1915120014.50231.00.9999
14.47013.4574130013.92981.00.9999
14.47013.7234140013.32121.00.9999
12.16263.9894150012.68141.00.9999
12.16264.2553160012.00991.00.9999
12.16264.5213170011.31791.00.9999
12.16264.7872180010.60171.00.9999
12.16265.053219009.88101.00.9999
9.51275.319120009.15671.00.9999
9.51275.585121008.44451.00.9999
9.51275.851122007.75731.00.9999
9.51276.117023007.10491.00.9999
9.51276.383024006.50161.00.9999
6.68736.648925005.95651.00.9999
6.68736.914926005.48531.00.9999
6.68737.180927005.09971.00.9999
6.68737.446828004.78891.00.9999
6.68737.712829004.55731.00.9999
4.74487.978730004.38891.00.9999
4.74488.244731004.26141.00.9999
4.74488.510632004.19601.00.9999
4.74488.776633004.13981.00.9999
4.74489.042634004.10921.00.9999
4.12539.308535004.09111.00.9999
4.12539.574536004.08511.00.9999
4.12539.840437004.07071.00.9999
4.125310.106438004.06301.00.9999
4.125310.372339004.05891.00.9999
4.039910.638340004.05741.00.9999
4.039910.904341004.04951.00.9999
4.039911.170242004.03671.00.9999
4.039911.436243004.02971.00.9999
4.039911.702144004.01681.00.9999
4.010211.968145004.00021.00.9999
4.010212.234046003.98231.00.9999
4.010212.547003.94741.00.9999
4.010212.766048003.88701.00.9999
4.010213.031949003.79331.00.9999
3.861613.297950003.65761.00.9999
3.861613.563851003.49251.00.9999
3.861613.829852003.25501.00.9999
3.861614.095753002.88361.00.8301
3.861614.361754002.52111.00.6171
3.02314.627755002.29021.00.5481
3.02314.893656002.10061.00.5079
3.02315.159657001.94641.00.4784
3.02315.425558001.81961.00.4597
3.02315.691559001.69751.00.4238
1.934815.957460001.60401.00.4093
1.934816.223461001.50351.00.4021
1.934816.489462001.42111.00.3930
1.934816.755363001.35291.00.3802
1.934817.021364001.27951.00.3791
1.412817.287265001.21931.00.3711
1.412817.553266001.16461.00.3674
1.412817.819167001.11931.00.3706
1.412818.085168001.06651.00.3606
1.412818.351169001.02440.99980.3590
1.101218.617070000.98641.00.3540
1.101218.883071000.95781.00.3554
1.101219.148972000.93090.99980.3509
1.101219.414973000.90701.00.3495
1.101219.680974000.86930.99980.3470
0.908319.946875000.84921.00.3449
0.908320.212876000.82141.00.3449
0.908320.478777000.82111.00.3500
0.908320.744778000.79641.00.3452
0.908321.010679000.77971.00.3429
0.754621.276680000.76341.00.3400
0.754621.542681000.74711.00.3384
0.754621.808582000.74001.00.3378
0.754622.074583000.72141.00.3390
0.754622.340484000.70620.99980.3375
0.65122.606485000.69731.00.3344
0.65122.872386000.69300.99980.3344
0.65123.138387000.68291.00.3350
0.65123.404388000.66831.00.3332
0.65123.670289000.65960.99980.3322
0.586823.936290000.67641.00.3321
0.586824.202191000.66350.99980.3308
0.586824.468192000.65601.00.3324
0.586824.734093000.64121.00.3290
0.586825.094000.63231.00.3307

Framework versions

  • —Transformers 4.47.0.dev0
  • —Pytorch 2.5.1+cu124
  • —Datasets 3.1.0
  • —Tokenizers 0.20.3