CoolFace
Modelpublic

oliverguhr/wav2vec2-base-german-cv9

sourceHugging Facemitupdated 4y agoView on Hugging Face
1likes178downloads
Model Card

wav2vec2-base-german-cv9

This model is a fine-tuned version of facebook/wav2vec2-base on the MOZILLA-FOUNDATION/COMMONVOICE9_0 - DE dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1742
  • Wer: 0.1209

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: 16
  • evalbatchsize: 32
  • seed: 42
  • gradientaccumulationsteps: 8
  • totaltrainbatch_size: 128
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lrschedulertype: linear
  • lrschedulerwarmup_ratio: 0.1
  • num_epochs: 50.0
  • mixedprecisiontraining: Native AMP

Training results

Training LossEpochStepValidation LossWer
0.68271.035570.66950.6247
0.39922.071140.37380.3936
0.26113.0106710.30110.3177
0.25364.0142280.26720.2749
0.19435.0177850.24870.2480
0.20046.0213420.22460.2268
0.16057.0248990.21760.2120
0.15798.0284560.20460.2024
0.16689.0320130.20270.1944
0.133810.0355700.19680.1854
0.147811.0391270.19630.1823
0.117712.0426840.19560.1800
0.124513.0462410.18890.1732
0.112414.0497980.18680.1714
0.111215.0533550.18050.1650
0.120916.0569120.18600.1614
0.100217.0604690.18280.1604
0.11818.0640260.18320.1580
0.097419.0675830.17710.1555
0.100720.0711400.18120.1532
0.086621.0746970.17520.1504
0.090122.0782540.16900.1477
0.096423.0818110.17730.1489
0.08524.0853680.17760.1456
0.094525.0889250.17860.1428
0.080426.0924820.17370.1429
0.083227.0960390.17890.1394
0.068328.0995960.17410.1390
0.076129.01031530.16880.1379
0.083330.01067100.17260.1370
0.075331.01102670.17740.1353
0.0832.01138240.17340.1344
0.064433.01173810.17370.1334
0.074534.01209380.17630.1335
0.062935.01244950.17610.1311
0.065436.01280520.17180.1302
0.065637.01316090.16970.1301
0.064338.01351660.17160.1279
0.068339.01387230.17770.1279
0.058740.01422800.17350.1271
0.069341.01458370.17800.1260
0.053242.01493940.17240.1245
0.059443.01529510.17360.1250
0.054444.01565080.17440.1238
0.055945.01600650.17700.1232
0.055746.01636220.17660.1231
0.052147.01671790.17510.1220
0.059148.01707360.17240.1217
0.050749.01742930.17530.1212
0.057750.01778500.17420.1209

Framework versions

  • Transformers 4.20.1
  • Pytorch 1.11.0+cu113
  • Datasets 2.0.0
  • Tokenizers 0.11.6