<!-- 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-timit-demo-colab971
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: 0.6551
- Wer: 0.4448
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: 2
- 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 |
---|---|---|---|---|
4.9461 | 1.77 | 500 | 3.2175 | 1.0 |
2.5387 | 3.53 | 1000 | 1.2239 | 0.7851 |
0.9632 | 5.3 | 1500 | 0.7275 | 0.6352 |
0.6585 | 7.07 | 2000 | 0.6218 | 0.5896 |
0.4875 | 8.83 | 2500 | 0.5670 | 0.5651 |
0.397 | 10.6 | 3000 | 0.5796 | 0.5487 |
0.3298 | 12.37 | 3500 | 0.5870 | 0.5322 |
0.2816 | 14.13 | 4000 | 0.5796 | 0.5016 |
0.2396 | 15.9 | 4500 | 0.5956 | 0.5040 |
0.2019 | 17.67 | 5000 | 0.5911 | 0.4847 |
0.1845 | 19.43 | 5500 | 0.6050 | 0.4800 |
0.1637 | 21.2 | 6000 | 0.6518 | 0.4927 |
0.1428 | 22.97 | 6500 | 0.6247 | 0.4645 |
0.1319 | 24.73 | 7000 | 0.6592 | 0.4711 |
0.1229 | 26.5 | 7500 | 0.6526 | 0.4556 |
0.1111 | 28.27 | 8000 | 0.6551 | 0.4448 |
Framework versions
- Transformers 4.11.3
- Pytorch 1.11.0+cu113
- Datasets 1.18.3
- Tokenizers 0.10.3