wav2vec2-large-xls-r-300m-gl-jupyter4
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.0970
- Wer: 0.0636
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: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 45
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 3.492 | 3.36 | 400 | 0.3109 | 0.3158 |
| 0.194 | 6.72 | 800 | 0.1279 | 0.1454 |
| 0.0794 | 10.08 | 1200 | 0.1210 | 0.1240 |
| 0.0565 | 13.44 | 1600 | 0.1209 | 0.1150 |
| 0.041 | 16.8 | 2000 | 0.1186 | 0.1107 |
| 0.0343 | 20.17 | 2400 | 0.1143 | 0.0933 |
| 0.0283 | 23.53 | 2800 | 0.1067 | 0.0900 |
| 0.0231 | 26.89 | 3200 | 0.1076 | 0.0812 |
| 0.0176 | 30.25 | 3600 | 0.1094 | 0.0780 |
| 0.0169 | 33.61 | 4000 | 0.1041 | 0.0766 |
| 0.0138 | 36.97 | 4400 | 0.1012 | 0.0711 |
| 0.0109 | 40.33 | 4800 | 0.0985 | 0.0655 |
| 0.0099 | 43.69 | 5200 | 0.0970 | 0.0636 |
Framework versions
- Transformers 4.11.3
- Pytorch 1.9.1+cu111
- Datasets 1.18.3
- Tokenizers 0.10.3