wav2vec2-xlsr-53-CV-demo-google-colab-Ezra_William_Prod15
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice_13_0 dataset. It achieves the following results on the evaluation set:
- Loss: 0.3344
- Wer: 0.2990
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: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 12
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
2.9553 | 1.0 | 278 | 2.9166 | 1.0 |
2.8419 | 2.0 | 556 | 2.1745 | 1.0 |
0.9699 | 3.0 | 834 | 0.5752 | 0.5677 |
0.6348 | 4.0 | 1112 | 0.4500 | 0.4575 |
0.5375 | 5.0 | 1390 | 0.3974 | 0.4070 |
0.4354 | 6.0 | 1668 | 0.3678 | 0.3576 |
0.3885 | 7.0 | 1946 | 0.3756 | 0.3539 |
0.3737 | 8.0 | 2224 | 0.3655 | 0.3345 |
0.336 | 9.0 | 2502 | 0.3472 | 0.3215 |
0.3014 | 10.0 | 2780 | 0.3395 | 0.3095 |
0.3103 | 11.0 | 3058 | 0.3311 | 0.3032 |
0.2965 | 12.0 | 3336 | 0.3344 | 0.2990 |
Framework versions
- Transformers 4.40.0
- Pytorch 2.2.2+cu121
- Datasets 2.19.0
- Tokenizers 0.19.1
- Downloads last month
- 15