<!-- 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. -->
thesis-audio-4
This model is a fine-tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.5585
- Wer: 0.3457
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: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 3.6041 | 1.0 | 500 | 2.7841 | 1.0 |
| 1.0447 | 2.01 | 1000 | 0.5261 | 0.5260 |
| 0.4404 | 3.01 | 1500 | 0.4699 | 0.4676 |
| 0.2945 | 4.02 | 2000 | 0.4232 | 0.4212 |
| 0.2223 | 5.02 | 2500 | 0.4348 | 0.4106 |
| 0.1849 | 6.02 | 3000 | 0.4559 | 0.4115 |
| 0.1566 | 7.03 | 3500 | 0.4942 | 0.3943 |
| 0.1389 | 8.03 | 4000 | 0.4142 | 0.3883 |
| 0.1244 | 9.04 | 4500 | 0.4382 | 0.3832 |
| 0.1028 | 10.04 | 5000 | 0.4644 | 0.3826 |
| 0.0972 | 11.04 | 5500 | 0.5119 | 0.3858 |
| 0.0868 | 12.05 | 6000 | 0.4886 | 0.3739 |
| 0.08 | 13.05 | 6500 | 0.5198 | 0.3736 |
| 0.0736 | 14.06 | 7000 | 0.4836 | 0.3672 |
| 0.0673 | 15.06 | 7500 | 0.5187 | 0.3769 |
| 0.0602 | 16.06 | 8000 | 0.6087 | 0.3800 |
| 0.0562 | 17.07 | 8500 | 0.5279 | 0.3630 |
| 0.0568 | 18.07 | 9000 | 0.5696 | 0.3700 |
| 0.047 | 19.08 | 9500 | 0.5964 | 0.3578 |
| 0.0426 | 20.08 | 10000 | 0.5801 | 0.3512 |
| 0.0411 | 21.08 | 10500 | 0.5889 | 0.3573 |
| 0.0349 | 22.09 | 11000 | 0.5654 | 0.3544 |
| 0.0342 | 23.09 | 11500 | 0.5610 | 0.3548 |
| 0.031 | 24.1 | 12000 | 0.5443 | 0.3468 |
| 0.0285 | 25.1 | 12500 | 0.5206 | 0.3469 |
| 0.0243 | 26.1 | 13000 | 0.5455 | 0.3484 |
| 0.0248 | 27.11 | 13500 | 0.5556 | 0.3474 |
| 0.0229 | 28.11 | 14000 | 0.5659 | 0.3457 |
| 0.0229 | 29.12 | 14500 | 0.5585 | 0.3457 |
Framework versions
- Transformers 4.17.0
- Pytorch 1.12.0+cu113
- Datasets 1.18.3
- Tokenizers 0.12.1
