metadata
license: apache-2.0
base_model: facebook/wav2vec2-base
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: w2v2-base-pretrained_lr5e-5_at0.8_da0.4
results: []
w2v2-base-pretrained_lr5e-5_at0.8_da0.4
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: 2.0085
- Wer: 0.1892
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: 5e-05
- 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: 500
- training_steps: 4000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 17.5459 | 13.16 | 250 | 3.5740 | 1.0 |
| 3.1637 | 26.32 | 500 | 3.2268 | 1.0 |
| 2.4912 | 39.47 | 750 | 1.4668 | 0.8590 |
| 0.2685 | 52.63 | 1000 | 1.2848 | 0.2469 |
| 0.0883 | 65.79 | 1250 | 1.4088 | 0.2302 |
| 0.0559 | 78.95 | 1500 | 1.5663 | 0.2196 |
| 0.0386 | 92.11 | 1750 | 1.7690 | 0.2063 |
| 0.0289 | 105.26 | 2000 | 1.7150 | 0.2114 |
| 0.0244 | 118.42 | 2250 | 1.8159 | 0.2106 |
| 0.018 | 131.58 | 2500 | 1.8855 | 0.2038 |
| 0.0147 | 144.74 | 2750 | 1.8758 | 0.2042 |
| 0.0121 | 157.89 | 3000 | 2.1326 | 0.2012 |
| 0.0105 | 171.05 | 3250 | 1.9975 | 0.1897 |
| 0.0093 | 184.21 | 3500 | 1.9094 | 0.1905 |
| 0.0079 | 197.37 | 3750 | 2.0668 | 0.1909 |
| 0.0082 | 210.53 | 4000 | 2.0085 | 0.1892 |
Framework versions
- Transformers 4.35.0
- Pytorch 2.0.0
- Datasets 2.14.6
- Tokenizers 0.14.1