metadata
license: apache-2.0
tags:
- automatic-speech-recognition
- /workspace/data/hy/noizy_student_3/
- generated_from_trainer
model-index:
- name: ''
results: []
This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on the /WORKSPACE/DATA/HY/NOIZY_STUDENT_3/ - NA dataset. It achieves the following results on the evaluation set:
- Loss: 0.1726
- Wer: 0.2260
- Cer: 0.0405
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: 16
- eval_batch_size: 64
- seed: 842
- gradient_accumulation_steps: 8
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.98) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- training_steps: 4000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Cer | Validation Loss | Wer |
|---|---|---|---|---|---|
| 1.331 | 7.02 | 400 | 0.0832 | 0.3281 | 0.4617 |
| 0.9968 | 14.04 | 800 | 0.0528 | 0.2114 | 0.3115 |
| 0.9367 | 21.05 | 1200 | 0.0481 | 0.1928 | 0.2771 |
| 0.9066 | 28.07 | 1600 | 0.0477 | 0.1905 | 0.2728 |
| 0.869 | 35.09 | 2000 | 0.0449 | 0.1817 | 0.2564 |
| 0.8319 | 42.11 | 2400 | 0.0433 | 0.1810 | 0.2490 |
| 0.8113 | 49.12 | 2800 | 0.0421 | 0.1769 | 0.2404 |
| 0.7624 | 56.14 | 3200 | 0.0432 | 0.1838 | 0.2443 |
| 0.7328 | 63.16 | 3600 | 0.0415 | 0.1756 | 0.2303 |
| 0.7209 | 70.18 | 4000 | 0.1726 | 0.2260 | 0.0405 |
Framework versions
- Transformers 4.17.0.dev0
- Pytorch 1.10.2
- Datasets 1.18.4.dev0
- Tokenizers 0.11.0