english-filipino-wav2vec2-l-xls-r-test-09
This model is a fine-tuned version of jonatasgrosman/wav2vec2-large-xlsr-53-english on the filipino_voice dataset. It achieves the following results on the evaluation set:
- Loss: 1.0054
- Wer: 0.5750
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: 0.002
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 20
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 3.001 | 2.09 | 400 | 1.3499 | 0.9595 |
| 0.8606 | 4.19 | 800 | 0.8159 | 0.6942 |
| 0.5819 | 6.28 | 1200 | 0.7372 | 0.6700 |
| 0.4751 | 8.38 | 1600 | 0.7310 | 0.6405 |
| 0.3777 | 10.47 | 2000 | 0.7841 | 0.6414 |
| 0.2918 | 12.57 | 2400 | 0.7898 | 0.5951 |
| 0.2209 | 14.66 | 2800 | 0.8558 | 0.5751 |
| 0.1671 | 16.75 | 3200 | 0.9881 | 0.5979 |
| 0.129 | 18.85 | 3600 | 1.0054 | 0.5750 |
Framework versions
- Transformers 4.11.3
- Pytorch 1.10.0+cu113
- Datasets 1.18.3
- Tokenizers 0.10.3