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_da0.7
results: []
w2v2-base-pretrained_lr5e-5_at0.8_da0.7
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: 1.7538
- 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 |
|---|---|---|---|---|
| 16.4265 | 7.81 | 250 | 3.9246 | 1.0 |
| 3.1637 | 15.62 | 500 | 3.1545 | 1.0 |
| 2.2078 | 23.44 | 750 | 1.2345 | 0.6258 |
| 0.2629 | 31.25 | 1000 | 1.0938 | 0.2862 |
| 0.1013 | 39.06 | 1250 | 1.2497 | 0.2482 |
| 0.061 | 46.88 | 1500 | 1.4009 | 0.1956 |
| 0.0438 | 54.69 | 1750 | 1.4359 | 0.1935 |
| 0.0334 | 62.5 | 2000 | 1.6005 | 0.2003 |
| 0.0256 | 70.31 | 2250 | 1.5475 | 0.1897 |
| 0.0231 | 78.12 | 2500 | 1.6901 | 0.1914 |
| 0.0176 | 85.94 | 2750 | 1.6743 | 0.1961 |
| 0.0164 | 93.75 | 3000 | 1.6498 | 0.1815 |
| 0.0133 | 101.56 | 3250 | 1.7255 | 0.1897 |
| 0.0127 | 109.38 | 3500 | 1.7270 | 0.1875 |
| 0.011 | 117.19 | 3750 | 1.7833 | 0.1918 |
| 0.0111 | 125.0 | 4000 | 1.7538 | 0.1892 |
Framework versions
- Transformers 4.35.0
- Pytorch 2.0.0
- Datasets 2.14.6
- Tokenizers 0.14.1