metadata
license: apache-2.0
base_model: jmaczan/wav2vec2-large-xls-r-300m-dysarthria
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: wav2vec2-large-xls-r-300m-dysarthria-big-dataset
results: []
wav2vec2-large-xls-r-300m-dysarthria-big-dataset
This model is a fine-tuned version of jmaczan/wav2vec2-large-xls-r-300m-dysarthria on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.0313
- Wer: 0.28
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: 0.0003
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 3.1965 | 1.6 | 200 | 1.7709 | 0.88 |
| 1.6135 | 3.2 | 400 | 1.1347 | 0.764 |
| 1.1769 | 4.8 | 600 | 0.9348 | 0.722 |
| 0.8155 | 6.4 | 800 | 0.5330 | 0.576 |
| 0.5961 | 8.0 | 1000 | 0.3927 | 0.504 |
| 0.4436 | 9.6 | 1200 | 0.3171 | 0.488 |
| 0.3581 | 11.2 | 1400 | 0.2877 | 0.512 |
| 0.2931 | 12.8 | 1600 | 0.1590 | 0.354 |
| 0.219 | 14.4 | 1800 | 0.1370 | 0.326 |
| 0.1912 | 16.0 | 2000 | 0.1362 | 0.262 |
| 0.1543 | 17.6 | 2200 | 0.0823 | 0.238 |
| 0.1323 | 19.2 | 2400 | 0.0764 | 0.272 |
| 0.1095 | 20.8 | 2600 | 0.0649 | 0.272 |
| 0.0909 | 22.4 | 2800 | 0.0537 | 0.274 |
| 0.0807 | 24.0 | 3000 | 0.0499 | 0.248 |
| 0.0618 | 25.6 | 3200 | 0.0499 | 0.318 |
| 0.0573 | 27.2 | 3400 | 0.0414 | 0.252 |
| 0.0456 | 28.8 | 3600 | 0.0313 | 0.28 |
Framework versions
- Transformers 4.43.2
- Pytorch 2.2.1+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1