metadata
license: apache-2.0
tags:
- generated_from_trainer
metrics:
- wer
base_model: facebook/wav2vec2-base
model-index:
- name: models
results: []
models
This model is a fine-tuned version of facebook/wav2vec2-base on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.0108
- Wer: 0.2638
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: 8
- 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: 1000
- num_epochs: 20
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 4.7187 | 1.26 | 500 | 2.7805 | 1.0 |
| 1.3764 | 2.53 | 1000 | 0.2018 | 0.4834 |
| 0.2713 | 3.79 | 1500 | 0.0572 | 0.3337 |
| 0.1924 | 5.05 | 2000 | 0.0486 | 0.3029 |
| 0.1283 | 6.31 | 2500 | 0.0254 | 0.2865 |
| 0.095 | 7.58 | 3000 | 0.0879 | 0.2924 |
| 0.0885 | 8.84 | 3500 | 0.0168 | 0.2702 |
| 0.0662 | 10.1 | 4000 | 0.0196 | 0.2708 |
| 0.0646 | 11.36 | 4500 | 0.0125 | 0.2697 |
| 0.0487 | 12.63 | 5000 | 0.0144 | 0.2691 |
| 0.0539 | 13.89 | 5500 | 0.0172 | 0.2673 |
| 0.0331 | 15.15 | 6000 | 0.0109 | 0.2656 |
| 0.0354 | 16.41 | 6500 | 0.0125 | 0.2656 |
| 0.0194 | 17.68 | 7000 | 0.0115 | 0.2650 |
| 0.0156 | 18.94 | 7500 | 0.0108 | 0.2638 |
Framework versions
- Transformers 4.39.3
- Pytorch 2.2.1+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2