wav2vec2-E50_feq_speed3
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: 0.9108
- Cer: 19.8766
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: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 50
- num_epochs: 3
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 36.8762 | 0.1289 | 200 | 4.9301 | 100.0 |
| 4.9098 | 0.2579 | 400 | 4.6288 | 100.0 |
| 4.7529 | 0.3868 | 600 | 4.6255 | 100.0 |
| 4.7332 | 0.5158 | 800 | 4.6055 | 100.0 |
| 4.6184 | 0.6447 | 1000 | 4.4703 | 100.0 |
| 4.0146 | 0.7737 | 1200 | 3.3130 | 54.5006 |
| 2.9541 | 0.9026 | 1400 | 2.6664 | 45.6522 |
| 2.5007 | 1.0316 | 1600 | 2.2122 | 38.3431 |
| 2.1884 | 1.1605 | 1800 | 1.9481 | 35.1469 |
| 1.9208 | 1.2895 | 2000 | 1.7513 | 32.5088 |
| 1.7158 | 1.4184 | 2200 | 1.7252 | 32.5617 |
| 1.5556 | 1.5474 | 2400 | 1.4556 | 29.0306 |
| 1.4658 | 1.6763 | 2600 | 1.4613 | 29.5828 |
| 1.3603 | 1.8053 | 2800 | 1.2933 | 26.1281 |
| 1.2334 | 1.9342 | 3000 | 1.2106 | 25.0294 |
| 1.1288 | 2.0632 | 3200 | 1.1409 | 23.7250 |
| 1.0335 | 2.1921 | 3400 | 1.0493 | 22.3854 |
| 0.9772 | 2.3211 | 3600 | 1.0525 | 22.1034 |
| 0.9549 | 2.4500 | 3800 | 0.9759 | 20.9342 |
| 0.907 | 2.5790 | 4000 | 0.9405 | 20.2350 |
| 0.8808 | 2.7079 | 4200 | 0.9395 | 20.2879 |
| 0.8458 | 2.8369 | 4400 | 0.9098 | 19.6886 |
| 0.8304 | 2.9658 | 4600 | 0.9108 | 19.8766 |
Framework versions
- Transformers 4.46.2
- Pytorch 2.5.1+cu121
- Datasets 3.1.0
- Tokenizers 0.20.3