|
--- |
|
license: apache-2.0 |
|
base_model: facebook/wav2vec2-xls-r-300m |
|
tags: |
|
- generated_from_trainer |
|
model-index: |
|
- name: wav2vec2-large-xls-r-korean-all |
|
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-large-xls-r-korean-all |
|
|
|
This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the None dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.1535 |
|
- Cer: 0.0329 |
|
|
|
## 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: 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: 10 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Cer | |
|
|:-------------:|:-----:|:-----:|:---------------:|:------:| |
|
| 3.0206 | 0.36 | 500 | 3.3589 | 0.9871 | |
|
| 0.6381 | 0.72 | 1000 | 0.6371 | 0.1714 | |
|
| 0.3951 | 1.08 | 1500 | 0.4320 | 0.125 | |
|
| 0.2858 | 1.44 | 2000 | 0.3546 | 0.1056 | |
|
| 0.2545 | 1.8 | 2500 | 0.2925 | 0.0872 | |
|
| 0.1833 | 2.16 | 3000 | 0.2520 | 0.0743 | |
|
| 0.1898 | 2.51 | 3500 | 0.2386 | 0.0679 | |
|
| 0.1981 | 2.87 | 4000 | 0.2135 | 0.0631 | |
|
| 0.123 | 3.23 | 4500 | 0.2129 | 0.0576 | |
|
| 0.1221 | 3.59 | 5000 | 0.2013 | 0.0543 | |
|
| 0.1218 | 3.95 | 5500 | 0.2000 | 0.0554 | |
|
| 0.1096 | 4.31 | 6000 | 0.1884 | 0.0507 | |
|
| 0.1113 | 4.67 | 6500 | 0.1781 | 0.0455 | |
|
| 0.075 | 5.03 | 7000 | 0.1811 | 0.0458 | |
|
| 0.0922 | 5.39 | 7500 | 0.1748 | 0.0455 | |
|
| 0.0766 | 5.75 | 8000 | 0.1807 | 0.0434 | |
|
| 0.0811 | 6.11 | 8500 | 0.1699 | 0.0411 | |
|
| 0.0876 | 6.47 | 9000 | 0.1641 | 0.0398 | |
|
| 0.0913 | 6.82 | 9500 | 0.1632 | 0.0392 | |
|
| 0.0658 | 7.18 | 10000 | 0.1667 | 0.0388 | |
|
| 0.0831 | 7.54 | 10500 | 0.1613 | 0.0375 | |
|
| 0.0716 | 7.9 | 11000 | 0.1552 | 0.0361 | |
|
| 0.0485 | 8.26 | 11500 | 0.1534 | 0.0351 | |
|
| 0.0469 | 8.62 | 12000 | 0.1541 | 0.0343 | |
|
| 0.0503 | 8.98 | 12500 | 0.1497 | 0.0340 | |
|
| 0.041 | 9.34 | 13000 | 0.1535 | 0.0337 | |
|
| 0.0556 | 9.7 | 13500 | 0.1535 | 0.0329 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.33.2 |
|
- Pytorch 1.12.1+cu113 |
|
- Datasets 2.14.5 |
|
- Tokenizers 0.13.3 |
|
|
