xls-r-300m-zv-mul
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.4882
- Wer: 0.4859
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: 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: 5
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 3.7438 | 0.26 | 500 | 0.9380 | 0.9135 |
| 1.1094 | 0.52 | 1000 | 0.5399 | 0.6874 |
| 0.9203 | 0.79 | 1500 | 0.5056 | 0.6708 |
| 0.8439 | 1.05 | 2000 | 0.4501 | 0.5775 |
| 0.7871 | 1.31 | 2500 | 0.4231 | 0.5592 |
| 0.761 | 1.57 | 3000 | 0.4335 | 0.5469 |
| 0.7309 | 1.83 | 3500 | 0.4204 | 0.5407 |
| 0.706 | 2.1 | 4000 | 0.4009 | 0.5177 |
| 0.6816 | 2.36 | 4500 | 0.3866 | 0.5108 |
| 0.6639 | 2.62 | 5000 | 0.3786 | 0.4895 |
| 0.6532 | 2.88 | 5500 | 0.3791 | 0.4895 |
| 0.6347 | 3.14 | 6000 | 0.3681 | 0.4740 |
| 0.6062 | 3.4 | 6500 | 0.3513 | 0.4695 |
| 0.5976 | 3.67 | 7000 | 0.3654 | 0.4779 |
| 0.5885 | 3.93 | 7500 | 0.3441 | 0.4552 |
| 0.5791 | 4.19 | 8000 | 0.3821 | 0.4610 |
| 0.6671 | 4.45 | 8500 | 0.4708 | 0.4981 |
| 0.6961 | 4.71 | 9000 | 0.4882 | 0.4859 |
Framework versions
- Transformers 4.26.1
- Pytorch 1.13.1+cu116
- Datasets 2.10.0
- Tokenizers 0.13.2