<!-- 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-base-demo-colab
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: 0.3944
- Wer: 0.3142
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: 32
- 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 |
|---|---|---|---|---|
| 3.4086 | 3.45 | 500 | 1.1494 | 0.8509 |
| 0.5968 | 6.9 | 1000 | 0.4306 | 0.4169 |
| 0.2363 | 10.34 | 1500 | 0.3820 | 0.3669 |
| 0.1365 | 13.79 | 2000 | 0.3863 | 0.3487 |
| 0.0916 | 17.24 | 2500 | 0.3851 | 0.3391 |
| 0.0704 | 20.69 | 3000 | 0.3759 | 0.3271 |
| 0.0537 | 24.14 | 3500 | 0.3747 | 0.3222 |
| 0.0413 | 27.59 | 4000 | 0.3944 | 0.3142 |
Framework versions
- Transformers 4.11.3
- Pytorch 1.11.0+cu113
- Datasets 1.14.0
- Tokenizers 0.10.3
