|
--- |
|
license: apache-2.0 |
|
tags: |
|
- generated_from_trainer |
|
model-index: |
|
- name: wav2vec2-russian |
|
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-russian |
|
|
|
This model is a fine-tuned version of [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on the None dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 1.2210 |
|
- Wer: 0.4966 |
|
|
|
## 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: 12 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Wer | |
|
|:-------------:|:-----:|:-----:|:---------------:|:------:| |
|
| 5.0548 | 0.25 | 500 | 4.1857 | 0.9999 | |
|
| 3.0204 | 0.5 | 1000 | 1.9996 | 0.9998 | |
|
| 1.8692 | 0.74 | 1500 | 1.6426 | 0.8698 | |
|
| 1.5154 | 0.99 | 2000 | 1.6156 | 0.7481 | |
|
| 1.3677 | 1.24 | 2500 | 2.1281 | 0.7120 | |
|
| 1.3223 | 1.49 | 3000 | 1.5192 | 0.6846 | |
|
| 1.2512 | 1.73 | 3500 | 1.0993 | 0.6634 | |
|
| 1.2257 | 1.98 | 4000 | 1.1039 | 0.6493 | |
|
| 1.1418 | 2.23 | 4500 | 1.0170 | 0.6241 | |
|
| 1.1213 | 2.48 | 5000 | 0.8436 | 0.6191 | |
|
| 1.112 | 2.73 | 5500 | 0.7326 | 0.6102 | |
|
| 1.0912 | 2.97 | 6000 | 0.7054 | 0.5976 | |
|
| 1.0465 | 3.22 | 6500 | 1.0887 | 0.5941 | |
|
| 1.0215 | 3.47 | 7000 | 1.4577 | 0.5793 | |
|
| 1.0244 | 3.72 | 7500 | 1.6058 | 0.5855 | |
|
| 1.0254 | 3.96 | 8000 | 1.3366 | 0.5750 | |
|
| 0.9558 | 4.21 | 8500 | 0.8088 | 0.5644 | |
|
| 0.966 | 4.46 | 9000 | 0.9650 | 0.5636 | |
|
| 0.9674 | 4.71 | 9500 | 0.9047 | 0.5532 | |
|
| 0.9373 | 4.96 | 10000 | 1.0342 | 0.5422 | |
|
| 0.9126 | 5.2 | 10500 | 1.2346 | 0.5462 | |
|
| 0.9063 | 5.45 | 11000 | 1.2696 | 0.5412 | |
|
| 0.9126 | 5.7 | 11500 | 1.4693 | 0.5317 | |
|
| 0.8936 | 5.95 | 12000 | 1.9096 | 0.5369 | |
|
| 0.8621 | 6.19 | 12500 | 1.6382 | 0.5326 | |
|
| 0.8695 | 6.44 | 13000 | 0.9466 | 0.5252 | |
|
| 0.8423 | 6.69 | 13500 | 1.6286 | 0.5355 | |
|
| 0.8494 | 6.94 | 14000 | 0.8368 | 0.5264 | |
|
| 0.8354 | 7.19 | 14500 | 0.6893 | 0.5216 | |
|
| 0.8133 | 7.43 | 15000 | 0.5916 | 0.5175 | |
|
| 0.8147 | 7.68 | 15500 | 0.7813 | 0.5221 | |
|
| 0.8258 | 7.93 | 16000 | 1.3814 | 0.5129 | |
|
| 0.8079 | 8.18 | 16500 | 0.8368 | 0.5176 | |
|
| 0.7868 | 8.42 | 17000 | 0.9456 | 0.5159 | |
|
| 0.7955 | 8.67 | 17500 | 0.7412 | 0.5170 | |
|
| 0.7921 | 8.92 | 18000 | 0.6256 | 0.5066 | |
|
| 0.7536 | 9.17 | 18500 | 0.8792 | 0.5101 | |
|
| 0.7667 | 9.42 | 19000 | 1.0615 | 0.5032 | |
|
| 0.772 | 9.66 | 19500 | 1.1312 | 0.5086 | |
|
| 0.7418 | 9.91 | 20000 | 1.3485 | 0.4990 | |
|
| 0.7577 | 10.16 | 20500 | 1.0788 | 0.5037 | |
|
| 0.7311 | 10.41 | 21000 | 0.9978 | 0.5032 | |
|
| 0.7419 | 10.65 | 21500 | 1.3925 | 0.5017 | |
|
| 0.74 | 10.9 | 22000 | 1.4191 | 0.4981 | |
|
| 0.7297 | 11.15 | 22500 | 1.1082 | 0.4994 | |
|
| 0.737 | 11.4 | 23000 | 1.1208 | 0.4971 | |
|
| 0.7266 | 11.65 | 23500 | 1.1595 | 0.4952 | |
|
| 0.7091 | 11.89 | 24000 | 1.2210 | 0.4966 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.17.0 |
|
- Pytorch 1.10.0+cu111 |
|
- Datasets 2.0.0 |
|
- Tokenizers 0.11.6 |
|
|
