Edit model card

wav2vec2-base-timit-demo-colab1

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

  • Loss: 3.1904
  • Wer: 1.0

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: 8
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 1000
  • num_epochs: 30
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
5.0877 1.42 500 3.2909 1.0
3.1333 2.85 1000 3.2624 1.0
3.1335 4.27 1500 3.2121 1.0
3.1294 5.7 2000 3.2047 1.0
3.1307 7.12 2500 3.2020 1.0
3.1279 8.55 3000 3.1978 1.0
3.1296 9.97 3500 3.2015 1.0
3.1273 11.4 4000 3.1983 1.0
3.1273 12.82 4500 3.2258 1.0
3.1274 14.25 5000 3.2151 1.0
3.1256 15.67 5500 3.2105 1.0
3.1302 17.09 6000 3.2018 1.0
3.1285 18.52 6500 3.2006 1.0
3.1251 19.94 7000 3.1858 1.0
3.1283 21.37 7500 3.1829 1.0
3.1267 22.79 8000 3.1773 1.0
3.1283 24.22 8500 3.1857 1.0
3.1253 25.64 9000 3.1847 1.0
3.1251 27.07 9500 3.1832 1.0
3.1245 28.49 10000 3.1869 1.0
3.1225 29.91 10500 3.1904 1.0

Framework versions

  • Transformers 4.11.3
  • Pytorch 1.11.0+cu113
  • Datasets 1.18.3
  • Tokenizers 0.10.3
Downloads last month
5