<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
wav2vec2-burak-new-300-v2-8-medium
This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.4311
- Wer: 0.2602
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: 0.0002
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 151
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 2.7459 | 9.43 | 500 | 0.3843 | 0.4597 |
| 0.579 | 18.87 | 1000 | 0.3006 | 0.3668 |
| 0.2662 | 28.3 | 1500 | 0.3760 | 0.3503 |
| 0.1936 | 37.74 | 2000 | 0.3631 | 0.3214 |
| 0.157 | 47.17 | 2500 | 0.3838 | 0.3063 |
| 0.1307 | 56.6 | 3000 | 0.3671 | 0.3056 |
| 0.1138 | 66.04 | 3500 | 0.3700 | 0.2959 |
| 0.1002 | 75.47 | 4000 | 0.4164 | 0.3014 |
| 0.0874 | 84.91 | 4500 | 0.4001 | 0.2973 |
| 0.0791 | 94.34 | 5000 | 0.3883 | 0.2911 |
| 0.0667 | 103.77 | 5500 | 0.4220 | 0.2780 |
| 0.0581 | 113.21 | 6000 | 0.4163 | 0.2670 |
| 0.0506 | 122.64 | 6500 | 0.4065 | 0.2753 |
| 0.043 | 132.08 | 7000 | 0.4279 | 0.2643 |
| 0.0386 | 141.51 | 7500 | 0.4284 | 0.2650 |
| 0.0341 | 150.94 | 8000 | 0.4311 | 0.2602 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.0+cu116
- Datasets 2.7.1
- Tokenizers 0.13.2
