TrainEsperanto
This model is a fine-tuned version of cpierse/wav2vec2-large-xlsr-53-esperanto on the audiofolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.0591
- Wer: 0.1884
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: 0.0003
- train_batch_size: 32
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 30
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 5.9902 | 2.6596 | 500 | 8.6294 | 1.0309 |
| 3.3 | 5.3191 | 1000 | 2.9688 | 1.0 |
| 2.8744 | 7.9787 | 1500 | 2.4117 | 1.0 |
| 0.7214 | 10.6383 | 2000 | 0.1825 | 0.2954 |
| 0.1552 | 13.2979 | 2500 | 0.0689 | 0.1971 |
| 0.1038 | 15.9574 | 3000 | 0.0621 | 0.1932 |
| 0.092 | 18.6170 | 3500 | 0.0624 | 0.1900 |
| 0.0877 | 21.2766 | 4000 | 0.0615 | 0.1926 |
| 0.082 | 23.9362 | 4500 | 0.0609 | 0.1899 |
| 0.0779 | 26.5957 | 5000 | 0.0591 | 0.1887 |
| 0.077 | 29.2553 | 5500 | 0.0591 | 0.1884 |
Framework versions
- Transformers 4.45.2
- Pytorch 2.5.1
- Datasets 2.19.1
- Tokenizers 0.20.1