metadata
license: apache-2.0
base_model: facebook/wav2vec2-base
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: w2v2-base-pretrained_lr5e-5_at0.6_da1
results: []
w2v2-base-pretrained_lr5e-5_at0.6_da1
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: 1.4542
- Wer: 0.1867
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 |
|---|---|---|---|---|
| 14.9488 | 4.24 | 250 | 3.6144 | 1.0 |
| 3.1468 | 8.47 | 500 | 3.2251 | 1.0 |
| 2.5166 | 12.71 | 750 | 1.3839 | 0.9603 |
| 0.4698 | 16.95 | 1000 | 0.6829 | 0.2909 |
| 0.2122 | 21.19 | 1250 | 0.9930 | 0.2217 |
| 0.1236 | 25.42 | 1500 | 1.1644 | 0.2140 |
| 0.0898 | 29.66 | 1750 | 1.0494 | 0.2076 |
| 0.0664 | 33.9 | 2000 | 1.1845 | 0.2093 |
| 0.0521 | 38.14 | 2250 | 1.2057 | 0.2089 |
| 0.0417 | 42.37 | 2500 | 1.3375 | 0.1914 |
| 0.0359 | 46.61 | 2750 | 1.5455 | 0.1880 |
| 0.0315 | 50.85 | 3000 | 1.3454 | 0.1884 |
| 0.0267 | 55.08 | 3250 | 1.2789 | 0.1944 |
| 0.0239 | 59.32 | 3500 | 1.3917 | 0.1909 |
| 0.0223 | 63.56 | 3750 | 1.4291 | 0.1897 |
| 0.0202 | 67.8 | 4000 | 1.4542 | 0.1867 |
Framework versions
- Transformers 4.35.0
- Pytorch 2.0.0
- Datasets 2.14.6
- Tokenizers 0.14.1