metadata
license: apache-2.0
base_model: facebook/wav2vec2-base
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: w2v2-base-pretrained_lr5e-5_at0.3_da1
results: []
w2v2-base-pretrained_lr5e-5_at0.3_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: 1.2363
- Wer: 0.1773
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 |
|---|---|---|---|---|
| 15.0228 | 3.97 | 250 | 4.0018 | 1.0 |
| 3.1757 | 7.94 | 500 | 3.2733 | 1.0 |
| 3.0888 | 11.9 | 750 | 3.0842 | 1.0 |
| 1.8583 | 15.87 | 1000 | 0.8173 | 0.6087 |
| 0.3832 | 19.84 | 1250 | 0.6533 | 0.3520 |
| 0.2105 | 23.81 | 1500 | 0.7165 | 0.2448 |
| 0.1301 | 27.78 | 1750 | 0.9040 | 0.2016 |
| 0.0917 | 31.75 | 2000 | 1.0631 | 0.1961 |
| 0.0674 | 35.71 | 2250 | 1.0216 | 0.1905 |
| 0.0571 | 39.68 | 2500 | 1.1088 | 0.1854 |
| 0.0456 | 43.65 | 2750 | 1.2525 | 0.1747 |
| 0.0387 | 47.62 | 3000 | 1.1236 | 0.1845 |
| 0.034 | 51.59 | 3250 | 1.1696 | 0.1730 |
| 0.0291 | 55.56 | 3500 | 1.2542 | 0.1756 |
| 0.0284 | 59.52 | 3750 | 1.2190 | 0.1760 |
| 0.026 | 63.49 | 4000 | 1.2363 | 0.1773 |
Framework versions
- Transformers 4.35.0
- Pytorch 2.0.0
- Datasets 2.14.6
- Tokenizers 0.14.1