<!-- 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-MIR_ST500-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: 2.7360
- Wer: 0.9837
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: 16
- 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: 500
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
101.0917 | 16.67 | 100 | 18.8979 | 0.8208 |
15.5054 | 33.33 | 200 | 10.9184 | 0.8208 |
10.1879 | 50.0 | 300 | 7.6480 | 0.8208 |
6.777 | 66.67 | 400 | 3.5386 | 1.0 |
3.0546 | 83.33 | 500 | 2.8794 | 1.0 |
2.8661 | 100.0 | 600 | 2.8405 | 1.0 |
2.847 | 116.67 | 700 | 2.8554 | 1.0 |
2.7661 | 133.33 | 800 | 2.6343 | 1.0 |
2.3474 | 150.0 | 900 | 2.7464 | 1.0 |
2.2464 | 166.67 | 1000 | 2.3565 | 1.0 |
2.207 | 183.33 | 1100 | 2.8854 | 1.0 |
2.3138 | 200.0 | 1200 | 2.5868 | 1.0 |
2.259 | 216.67 | 1300 | 2.6530 | 1.0 |
2.1667 | 233.33 | 1400 | 2.4921 | 1.0 |
2.1268 | 250.0 | 1500 | 2.5435 | 1.0 |
2.1089 | 266.67 | 1600 | 2.5444 | 1.0 |
2.0845 | 283.33 | 1700 | 2.6796 | 1.0 |
2.0672 | 300.0 | 1800 | 2.5824 | 1.0 |
2.055 | 316.67 | 1900 | 2.4631 | 1.0 |
2.0317 | 333.33 | 2000 | 2.5751 | 1.0 |
2.0141 | 350.0 | 2100 | 2.5627 | 1.0 |
1.9914 | 366.67 | 2200 | 2.6132 | 1.0 |
1.9489 | 383.33 | 2300 | 2.7527 | 1.0 |
1.9146 | 400.0 | 2400 | 2.6121 | 0.9935 |
1.893 | 416.67 | 2500 | 2.7110 | 0.9902 |
1.845 | 433.33 | 2600 | 2.7410 | 0.9967 |
1.8095 | 450.0 | 2700 | 2.7013 | 0.9935 |
1.7708 | 466.67 | 2800 | 2.7719 | 0.9935 |
1.7224 | 483.33 | 2900 | 2.7740 | 0.9837 |
1.6961 | 500.0 | 3000 | 2.7360 | 0.9837 |
Framework versions
- Transformers 4.11.3
- Pytorch 1.9.1+cu102
- Datasets 1.18.3
- Tokenizers 0.10.3