Edit model card

ai-light-dance_singing6_ft_wav2vec2-large-xlsr-53-5gram-v4-2

This model is a fine-tuned version of gary109/ai-light-dance_singing6_ft_wav2vec2-large-xlsr-53-5gram-v4-2 on the GARY109/AI_LIGHT_DANCE - ONSET-SINGING6 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1731
  • Wer: 0.0949

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: 4e-06
  • train_batch_size: 8
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 8
  • total_train_batch_size: 64
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 100
  • num_epochs: 50.0
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
0.4688 1.0 171 0.1822 0.0986
0.4505 2.0 342 0.1822 0.0995
0.4418 3.0 513 0.1819 0.0996
0.4505 4.0 684 0.1842 0.1009
0.4403 5.0 855 0.1817 0.0983
0.4362 6.0 1026 0.1820 0.1001
0.4546 7.0 1197 0.1795 0.0979
0.4283 8.0 1368 0.1841 0.1004
0.4289 9.0 1539 0.1783 0.0970
0.445 10.0 1710 0.1785 0.0969
0.4483 11.0 1881 0.1781 0.0990
0.4319 12.0 2052 0.1811 0.0995
0.4175 13.0 2223 0.1781 0.0979
0.4298 14.0 2394 0.1761 0.0969
0.4511 15.0 2565 0.1741 0.0950
0.4287 16.0 2736 0.1771 0.0976
0.4417 17.0 2907 0.1780 0.0979
0.4257 18.0 3078 0.1762 0.0963
0.4588 19.0 3249 0.1750 0.0957
0.4439 20.0 3420 0.1758 0.0968
0.4515 21.0 3591 0.1782 0.0966
0.4311 22.0 3762 0.1774 0.0975
0.403 23.0 3933 0.1758 0.0963
0.4168 24.0 4104 0.1775 0.0972
0.425 25.0 4275 0.1742 0.0952
0.4493 26.0 4446 0.1749 0.0963
0.4232 27.0 4617 0.1749 0.0966
0.4331 28.0 4788 0.1754 0.0964
0.4306 29.0 4959 0.1756 0.0967
0.4261 30.0 5130 0.1753 0.0969
0.4284 31.0 5301 0.1749 0.0958
0.4322 32.0 5472 0.1748 0.0952
0.4225 33.0 5643 0.1747 0.0952
0.4179 34.0 5814 0.1749 0.0955
0.4264 35.0 5985 0.1757 0.0966
0.4217 36.0 6156 0.1753 0.0955
0.4556 37.0 6327 0.1749 0.0957
0.4181 38.0 6498 0.1756 0.0972
0.4286 39.0 6669 0.1747 0.0956
0.4427 40.0 6840 0.1747 0.0965
0.4292 41.0 7011 0.1742 0.0955
0.424 42.0 7182 0.1740 0.0952
0.4314 43.0 7353 0.1743 0.0963
0.4359 44.0 7524 0.1744 0.0952
0.4195 45.0 7695 0.1736 0.0949
0.4214 46.0 7866 0.1731 0.0949
0.4358 47.0 8037 0.1738 0.0952
0.4347 48.0 8208 0.1742 0.0956
0.4032 49.0 8379 0.1739 0.0956
0.441 50.0 8550 0.1737 0.0957

Framework versions

  • Transformers 4.21.0.dev0
  • Pytorch 1.9.1+cu102
  • Datasets 2.3.3.dev0
  • Tokenizers 0.12.1
Downloads last month
67
Hosted inference API
or or
This model can be loaded on the Inference API on-demand.