wav2vec2-english-learner-google-colab
This model is a fine-tuned version of facebook/wav2vec2-base on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.6979
- Wer: 0.5039
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: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- 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 |
|---|---|---|---|---|
| 10.3630 | 1.3055 | 500 | 6.1820 | 1.0004 |
| 5.0714 | 2.6110 | 1000 | 5.0490 | 1.0029 |
| 3.5323 | 3.9164 | 1500 | 1.6699 | 1.0502 |
| 1.3650 | 5.2219 | 2000 | 1.0074 | 0.9129 |
| 0.8832 | 6.5274 | 2500 | 0.9509 | 0.7993 |
| 0.6205 | 7.8329 | 3000 | 0.7623 | 0.7204 |
| 0.4967 | 9.1384 | 3500 | 0.7962 | 0.6699 |
| 0.3820 | 10.4439 | 4000 | 0.6703 | 0.6545 |
| 0.3197 | 11.7493 | 4500 | 0.6283 | 0.6097 |
| 0.2523 | 13.0548 | 5000 | 0.5867 | 0.5853 |
| 0.2085 | 14.3603 | 5500 | 0.6501 | 0.5667 |
| 0.1768 | 15.6658 | 6000 | 0.6836 | 0.5530 |
| 0.1598 | 16.9713 | 6500 | 0.6432 | 0.5563 |
| 0.1431 | 18.2768 | 7000 | 0.6522 | 0.5401 |
| 0.1230 | 19.5822 | 7500 | 0.6357 | 0.5323 |
| 0.1064 | 20.8877 | 8000 | 0.6997 | 0.5301 |
| 0.0931 | 22.1932 | 8500 | 0.6959 | 0.5290 |
| 0.0809 | 23.4987 | 9000 | 0.7457 | 0.5208 |
| 0.0765 | 24.8042 | 9500 | 0.7360 | 0.5154 |
| 0.0637 | 26.1097 | 10000 | 0.7098 | 0.5115 |
| 0.0604 | 27.4151 | 10500 | 0.7237 | 0.5047 |
| 0.0558 | 28.7206 | 11000 | 0.6887 | 0.5043 |
| 0.0558 | 30.0 | 11490 | 0.6979 | 0.5039 |
Framework versions
- Transformers 5.12.1
- Pytorch 2.11.0+cu128
- Datasets 5.0.0
- Tokenizers 0.22.2