wav2vec2-xls-r-1b-mucs-62255-hinglish_mixed_scripts-25_45-1e-4-epochs-100-FT
This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.7631
- Global Wer: 25.8799
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: 16
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: constant_with_warmup
- lr_scheduler_warmup_steps: 500
- num_epochs: 100
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Global Wer |
|---|---|---|---|---|
| No log | 1.0 | 440 | 1.2057 | 58.0255 |
| 5.7157 | 2.0 | 880 | 0.9840 | 45.5704 |
| 1.5197 | 3.0 | 1320 | 0.8813 | 30.8924 |
| 1.1859 | 4.0 | 1760 | 0.8179 | 28.8330 |
| 1.0175 | 5.0 | 2200 | 0.7491 | 27.7869 |
| 0.8691 | 6.0 | 2640 | 0.7963 | 27.8958 |
| 0.7622 | 7.0 | 3080 | 0.7655 | 28.7131 |
| 0.6768 | 8.0 | 3520 | 0.7797 | 25.4440 |
| 0.6768 | 9.0 | 3960 | 0.7329 | 26.5773 |
| 0.5882 | 10.0 | 4400 | 0.7663 | 26.1850 |
| 0.5132 | 11.0 | 4840 | 0.7631 | 25.8799 |
Framework versions
- Transformers 5.13.0
- Pytorch 2.6.0+cu124
- Datasets 3.6.0
- Tokenizers 0.22.2