|
--- |
|
license: cc-by-nc-4.0 |
|
base_model: nguyenvulebinh/wav2vec2-base-vi |
|
tags: |
|
- generated_from_trainer |
|
metrics: |
|
- wer |
|
model-index: |
|
- name: wav2vec2_base_vietnamese_control_dataset |
|
results: [] |
|
--- |
|
|
|
<!-- This model card has been generated automatically according to the information the Trainer had access to. You |
|
should probably proofread and complete it, then remove this comment. --> |
|
|
|
# wav2vec2_base_vietnamese_control_dataset |
|
|
|
This model is a fine-tuned version of [nguyenvulebinh/wav2vec2-base-vi](https://huggingface.co/nguyenvulebinh/wav2vec2-base-vi) on an unknown dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.0491 |
|
- Wer: 0.2007 |
|
|
|
## 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: 1e-05 |
|
- train_batch_size: 16 |
|
- 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: 100 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Wer | |
|
|:-------------:|:-----:|:-----:|:---------------:|:------:| |
|
| 19.8101 | 2.05 | 500 | 17.9020 | 1.0 | |
|
| 12.7254 | 4.1 | 1000 | 13.4585 | 1.0 | |
|
| 8.6927 | 6.15 | 1500 | 8.4217 | 1.0 | |
|
| 5.6269 | 8.2 | 2000 | 5.1617 | 1.0 | |
|
| 3.9616 | 10.25 | 2500 | 3.6167 | 1.0 | |
|
| 3.2881 | 12.3 | 3000 | 3.2056 | 1.0 | |
|
| 3.1381 | 14.34 | 3500 | 3.1174 | 1.0 | |
|
| 3.092 | 16.39 | 4000 | 3.1044 | 1.0 | |
|
| 3.0222 | 18.44 | 4500 | 2.9461 | 1.0 | |
|
| 2.8394 | 20.49 | 5000 | 2.7133 | 1.0 | |
|
| 2.6096 | 22.54 | 5500 | 2.3871 | 1.0 | |
|
| 2.3244 | 24.59 | 6000 | 2.0311 | 1.0 | |
|
| 1.9785 | 26.64 | 6500 | 1.6117 | 1.0 | |
|
| 1.6346 | 28.69 | 7000 | 1.2173 | 1.0 | |
|
| 1.3122 | 30.74 | 7500 | 0.9547 | 1.0 | |
|
| 1.0927 | 32.79 | 8000 | 0.7738 | 1.0 | |
|
| 0.9261 | 34.84 | 8500 | 0.6172 | 0.8970 | |
|
| 0.7336 | 36.89 | 9000 | 0.4357 | 0.3654 | |
|
| 0.5754 | 38.93 | 9500 | 0.3304 | 0.3071 | |
|
| 0.4791 | 40.98 | 10000 | 0.2668 | 0.2785 | |
|
| 0.4212 | 43.03 | 10500 | 0.2240 | 0.2548 | |
|
| 0.3439 | 45.08 | 11000 | 0.1852 | 0.2329 | |
|
| 0.3048 | 47.13 | 11500 | 0.1607 | 0.2119 | |
|
| 0.2684 | 49.18 | 12000 | 0.1376 | 0.2105 | |
|
| 0.2298 | 51.23 | 12500 | 0.1227 | 0.2071 | |
|
| 0.2192 | 53.28 | 13000 | 0.1092 | 0.2055 | |
|
| 0.2063 | 55.33 | 13500 | 0.0990 | 0.2039 | |
|
| 0.1875 | 57.38 | 14000 | 0.0895 | 0.2039 | |
|
| 0.1692 | 59.43 | 14500 | 0.0822 | 0.2039 | |
|
| 0.1421 | 61.48 | 15000 | 0.0766 | 0.2029 | |
|
| 0.1505 | 63.52 | 15500 | 0.0710 | 0.2031 | |
|
| 0.1796 | 65.57 | 16000 | 0.0682 | 0.2019 | |
|
| 0.1265 | 67.62 | 16500 | 0.0641 | 0.2015 | |
|
| 0.1172 | 69.67 | 17000 | 0.0617 | 0.2019 | |
|
| 0.1173 | 71.72 | 17500 | 0.0586 | 0.2011 | |
|
| 0.1226 | 73.77 | 18000 | 0.0568 | 0.2015 | |
|
| 0.1165 | 75.82 | 18500 | 0.0567 | 0.2011 | |
|
| 0.1098 | 77.87 | 19000 | 0.0547 | 0.2007 | |
|
| 0.0996 | 79.92 | 19500 | 0.0537 | 0.2009 | |
|
| 0.1024 | 81.97 | 20000 | 0.0521 | 0.2009 | |
|
| 0.0992 | 84.02 | 20500 | 0.0508 | 0.2009 | |
|
| 0.1008 | 86.07 | 21000 | 0.0510 | 0.2009 | |
|
| 0.1147 | 88.11 | 21500 | 0.0501 | 0.2007 | |
|
| 0.1138 | 90.16 | 22000 | 0.0500 | 0.2005 | |
|
| 0.0939 | 92.21 | 22500 | 0.0493 | 0.2007 | |
|
| 0.1021 | 94.26 | 23000 | 0.0492 | 0.2005 | |
|
| 0.1009 | 96.31 | 23500 | 0.0488 | 0.2009 | |
|
| 0.0935 | 98.36 | 24000 | 0.0491 | 0.2007 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.32.1 |
|
- Pytorch 2.0.1+cu117 |
|
- Datasets 2.14.4 |
|
- Tokenizers 0.13.3 |
|
|
