Hugging Face's logo

Finnish-NLP
/
wav2vec2-large-uralic-voxpopuli-v2-finnish

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
Finnish
wav2vec2
finnish
Generated from Trainer
hf-asr-leaderboard
Eval Results
Inference Endpoints
Model card Files Metrics Community
wav2vec2-large-uralic-voxpopuli-v2-finnish / README.md
aapot's picture
aapot
update model card README.md
fff5d98
preview code
|
raw
history blame
5.08 kB
metadata
license: cc-by-nc-4.0
tags:
  - generated_from_trainer
model-index:
  - name: wav2vec2-large-uralic-voxpopuli-v2-finnish
    results: []

wav2vec2-large-uralic-voxpopuli-v2-finnish

This model is a fine-tuned version of facebook/wav2vec2-large-uralic-voxpopuli-v2 on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0828
  • Wer: 0.1075

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
  • train_batch_size: 32
  • eval_batch_size: 32
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 500
  • num_epochs: 10
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
1.9421 0.17 500 0.8633 0.8870
0.572 0.33 1000 0.1650 0.1829
0.5149 0.5 1500 0.1416 0.1711
0.4884 0.66 2000 0.1265 0.1605
0.4729 0.83 2500 0.1205 0.1485
0.4723 1.0 3000 0.1108 0.1403
0.443 1.16 3500 0.1175 0.1439
0.4378 1.33 4000 0.1083 0.1482
0.4313 1.49 4500 0.1110 0.1398
0.4182 1.66 5000 0.1024 0.1418
0.3884 1.83 5500 0.1032 0.1395
0.4034 1.99 6000 0.0985 0.1318
0.3735 2.16 6500 0.1008 0.1355
0.4174 2.32 7000 0.0970 0.1361
0.3581 2.49 7500 0.0968 0.1297
0.3783 2.66 8000 0.0881 0.1284
0.3827 2.82 8500 0.0921 0.1352
0.3651 2.99 9000 0.0861 0.1298
0.3684 3.15 9500 0.0844 0.1270
0.3784 3.32 10000 0.0870 0.1248
0.356 3.48 10500 0.0828 0.1214
0.3524 3.65 11000 0.0878 0.1218
0.3879 3.82 11500 0.0874 0.1216
0.3521 3.98 12000 0.0860 0.1210
0.3527 4.15 12500 0.0818 0.1184
0.3529 4.31 13000 0.0787 0.1185
0.3114 4.48 13500 0.0852 0.1202
0.3495 4.65 14000 0.0807 0.1187
0.34 4.81 14500 0.0796 0.1162
0.3646 4.98 15000 0.0782 0.1149
0.3004 5.14 15500 0.0799 0.1142
0.3167 5.31 16000 0.0847 0.1123
0.3249 5.48 16500 0.0837 0.1171
0.3202 5.64 17000 0.0749 0.1109
0.3104 5.81 17500 0.0798 0.1093
0.3039 5.97 18000 0.0810 0.1132
0.3157 6.14 18500 0.0847 0.1156
0.3133 6.31 19000 0.0833 0.1140
0.3203 6.47 19500 0.0838 0.1113
0.3178 6.64 20000 0.0907 0.1141
0.3182 6.8 20500 0.0938 0.1143
0.3 6.97 21000 0.0854 0.1133
0.3151 7.14 21500 0.0859 0.1109
0.2963 7.3 22000 0.0832 0.1122
0.3099 7.47 22500 0.0865 0.1103
0.322 7.63 23000 0.0833 0.1105
0.3064 7.8 23500 0.0865 0.1078
0.2964 7.97 24000 0.0859 0.1096
0.2869 8.13 24500 0.0872 0.1100
0.315 8.3 25000 0.0869 0.1099
0.3003 8.46 25500 0.0878 0.1105
0.2947 8.63 26000 0.0884 0.1084
0.297 8.8 26500 0.0891 0.1102
0.3049 8.96 27000 0.0863 0.1081
0.2957 9.13 27500 0.0846 0.1083
0.2908 9.29 28000 0.0848 0.1059
0.2955 9.46 28500 0.0846 0.1085
0.2991 9.62 29000 0.0839 0.1081
0.3112 9.79 29500 0.0832 0.1071
0.29 9.96 30000 0.0828 0.1075

Framework versions

  • Transformers 4.19.2
  • Pytorch 1.11.0+cu102
  • Datasets 2.2.2
  • Tokenizers 0.11.0