|
--- |
|
license: apache-2.0 |
|
tags: |
|
- generated_from_trainer |
|
datasets: |
|
- common_voice |
|
model-index: |
|
- name: wav2vec2-xls-r-300m-gn-cv8 |
|
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-xls-r-300m-gn-cv8 |
|
|
|
This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the common_voice dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 1.0820 |
|
- Wer: 0.7212 |
|
|
|
## 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: 1 |
|
- eval_batch_size: 8 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 8 |
|
- total_train_batch_size: 8 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- lr_scheduler_warmup_steps: 100 |
|
- training_steps: 5000 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Wer | |
|
|:-------------:|:------:|:----:|:---------------:|:------:| |
|
| 12.7853 | 2.76 | 100 | 4.7861 | 1.0 | |
|
| 3.4153 | 5.55 | 200 | 3.5519 | 1.0 | |
|
| 3.2923 | 8.33 | 300 | 3.3052 | 1.0 | |
|
| 3.2119 | 11.11 | 400 | 3.1202 | 1.0 | |
|
| 2.5099 | 13.87 | 500 | 1.6023 | 0.9872 | |
|
| 1.3373 | 16.66 | 600 | 1.1878 | 0.9182 | |
|
| 0.913 | 19.44 | 700 | 1.0049 | 0.8875 | |
|
| 0.7013 | 22.22 | 800 | 0.9810 | 0.8542 | |
|
| 0.5439 | 24.98 | 900 | 0.9463 | 0.8568 | |
|
| 0.4581 | 27.76 | 1000 | 0.9771 | 0.8261 | |
|
| 0.392 | 30.55 | 1100 | 0.9489 | 0.8389 | |
|
| 0.3555 | 33.33 | 1200 | 0.8846 | 0.8107 | |
|
| 0.3219 | 36.11 | 1300 | 0.8567 | 0.7980 | |
|
| 0.2794 | 38.87 | 1400 | 0.8851 | 0.7775 | |
|
| 0.2649 | 41.66 | 1500 | 0.9642 | 0.7954 | |
|
| 0.2407 | 44.44 | 1600 | 0.9540 | 0.8133 | |
|
| 0.2184 | 47.22 | 1700 | 0.8820 | 0.7494 | |
|
| 0.2181 | 49.98 | 1800 | 0.9349 | 0.8031 | |
|
| 0.1863 | 52.76 | 1900 | 0.9557 | 0.7494 | |
|
| 0.1728 | 55.55 | 2000 | 1.0587 | 0.7519 | |
|
| 0.1848 | 58.33 | 2100 | 1.0072 | 0.8056 | |
|
| 0.1602 | 61.11 | 2200 | 0.9321 | 0.7980 | |
|
| 0.1479 | 63.87 | 2300 | 0.9669 | 0.8005 | |
|
| 0.1464 | 66.66 | 2400 | 0.9914 | 0.7545 | |
|
| 0.1442 | 69.44 | 2500 | 1.0479 | 0.8184 | |
|
| 0.1385 | 72.22 | 2600 | 1.0065 | 0.7647 | |
|
| 0.1201 | 74.98 | 2700 | 0.9956 | 0.7801 | |
|
| 0.1264 | 77.76 | 2800 | 1.0153 | 0.7801 | |
|
| 0.1143 | 80.55 | 2900 | 0.9973 | 0.7826 | |
|
| 0.1145 | 83.33 | 3000 | 0.9762 | 0.7698 | |
|
| 0.1264 | 86.11 | 3100 | 0.9494 | 0.7391 | |
|
| 0.1093 | 88.87 | 3200 | 1.0091 | 0.7801 | |
|
| 0.0988 | 91.66 | 3300 | 1.0605 | 0.7621 | |
|
| 0.103 | 94.44 | 3400 | 0.9910 | 0.7340 | |
|
| 0.0972 | 97.22 | 3500 | 1.0412 | 0.7519 | |
|
| 0.0974 | 99.98 | 3600 | 1.0361 | 0.7621 | |
|
| 0.0836 | 102.76 | 3700 | 0.9969 | 0.7673 | |
|
| 0.0795 | 105.55 | 3800 | 1.0198 | 0.7545 | |
|
| 0.0839 | 108.33 | 3900 | 1.0269 | 0.7698 | |
|
| 0.0856 | 111.11 | 4000 | 0.9913 | 0.7442 | |
|
| 0.0721 | 113.87 | 4100 | 1.0239 | 0.7621 | |
|
| 0.0711 | 116.66 | 4200 | 1.0360 | 0.7468 | |
|
| 0.0771 | 119.44 | 4300 | 1.0799 | 0.7289 | |
|
| 0.0624 | 122.22 | 4400 | 1.1323 | 0.7238 | |
|
| 0.0748 | 124.98 | 4500 | 1.0868 | 0.7366 | |
|
| 0.0644 | 127.76 | 4600 | 1.0658 | 0.7289 | |
|
| 0.0667 | 130.55 | 4700 | 1.0731 | 0.7212 | |
|
| 0.0624 | 133.33 | 4800 | 1.0794 | 0.7289 | |
|
| 0.0714 | 136.11 | 4900 | 1.0832 | 0.7238 | |
|
| 0.0627 | 138.87 | 5000 | 1.0820 | 0.7212 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.15.0 |
|
- Pytorch 1.10.0+cu111 |
|
- Datasets 1.18.1 |
|
- Tokenizers 0.10.3 |
|
|