XLS-R-demo-google-colab-Ezra_William_Prod_2
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice_13_0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.8224
- Wer: 0.7215
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: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 12
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 5.1374 | 1.0 | 121 | 2.9882 | 1.0 |
| 2.9639 | 2.0 | 242 | 2.9334 | 1.0 |
| 2.9322 | 3.0 | 363 | 2.9144 | 1.0 |
| 2.9198 | 4.0 | 484 | 2.9017 | 1.0 |
| 2.8947 | 5.0 | 605 | 2.8740 | 1.0 |
| 2.8716 | 6.0 | 726 | 2.8349 | 0.9999 |
| 2.8277 | 7.0 | 847 | 2.5051 | 1.0 |
| 2.5331 | 8.0 | 968 | 1.4294 | 0.9178 |
| 1.6883 | 9.0 | 1089 | 0.9830 | 0.7981 |
| 0.9526 | 10.0 | 1210 | 0.8735 | 0.7564 |
| 0.8469 | 11.0 | 1331 | 0.8294 | 0.7308 |
| 0.83 | 12.0 | 1452 | 0.8224 | 0.7215 |
Framework versions
- Transformers 4.42.3
- Pytorch 2.3.1+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1