<!-- 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-xlsr-mecita-coraa-portuguese-all-clean-04
This model is a fine-tuned version of Edresson/wav2vec2-large-xlsr-coraa-portuguese on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.1664
- Wer: 0.0944
- Cer: 0.0281
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: 3e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 100
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
---|---|---|---|---|---|
21.1889 | 1.0 | 67 | 3.3270 | 1.0 | 1.0 |
7.4773 | 2.0 | 134 | 2.9699 | 1.0 | 1.0 |
3.0098 | 3.0 | 201 | 2.9123 | 1.0 | 1.0 |
3.0098 | 4.0 | 268 | 2.9135 | 1.0 | 1.0 |
2.9096 | 5.0 | 335 | 2.8647 | 1.0 | 1.0 |
2.8469 | 6.0 | 402 | 2.4765 | 1.0 | 0.9998 |
2.8469 | 7.0 | 469 | 0.8785 | 0.4283 | 0.1059 |
1.7225 | 8.0 | 536 | 0.4984 | 0.2262 | 0.0588 |
0.8041 | 9.0 | 603 | 0.3737 | 0.1973 | 0.0517 |
0.8041 | 10.0 | 670 | 0.3288 | 0.1816 | 0.0496 |
0.5862 | 11.0 | 737 | 0.3055 | 0.1751 | 0.0474 |
0.4768 | 12.0 | 804 | 0.2713 | 0.1492 | 0.0411 |
0.4768 | 13.0 | 871 | 0.2496 | 0.1394 | 0.0393 |
0.4019 | 14.0 | 938 | 0.2453 | 0.1319 | 0.0380 |
0.3483 | 15.0 | 1005 | 0.2310 | 0.1203 | 0.0353 |
0.3483 | 16.0 | 1072 | 0.2185 | 0.1196 | 0.0347 |
0.3276 | 17.0 | 1139 | 0.2147 | 0.1152 | 0.0345 |
0.2945 | 18.0 | 1206 | 0.2067 | 0.1090 | 0.0324 |
0.2945 | 19.0 | 1273 | 0.1995 | 0.1066 | 0.0327 |
0.3026 | 20.0 | 1340 | 0.2011 | 0.1019 | 0.0322 |
0.2541 | 21.0 | 1407 | 0.1959 | 0.1022 | 0.0316 |
0.2541 | 22.0 | 1474 | 0.1960 | 0.1046 | 0.0318 |
0.2472 | 23.0 | 1541 | 0.1890 | 0.1056 | 0.0321 |
0.2374 | 24.0 | 1608 | 0.1882 | 0.0971 | 0.0311 |
0.2374 | 25.0 | 1675 | 0.1843 | 0.0961 | 0.0300 |
0.2217 | 26.0 | 1742 | 0.1837 | 0.1032 | 0.0311 |
0.2352 | 27.0 | 1809 | 0.1845 | 0.0971 | 0.0304 |
0.2352 | 28.0 | 1876 | 0.1809 | 0.0985 | 0.0303 |
0.2207 | 29.0 | 1943 | 0.1761 | 0.0937 | 0.0297 |
0.2067 | 30.0 | 2010 | 0.1791 | 0.0961 | 0.0294 |
0.2067 | 31.0 | 2077 | 0.1764 | 0.0978 | 0.0299 |
0.1956 | 32.0 | 2144 | 0.1762 | 0.1019 | 0.0303 |
0.1981 | 33.0 | 2211 | 0.1809 | 0.0964 | 0.0299 |
0.1981 | 34.0 | 2278 | 0.1828 | 0.0944 | 0.0298 |
0.1741 | 35.0 | 2345 | 0.1799 | 0.0930 | 0.0293 |
0.1776 | 36.0 | 2412 | 0.1748 | 0.0930 | 0.0289 |
0.1776 | 37.0 | 2479 | 0.1767 | 0.0927 | 0.0292 |
0.184 | 38.0 | 2546 | 0.1755 | 0.0882 | 0.0286 |
0.1887 | 39.0 | 2613 | 0.1753 | 0.0927 | 0.0292 |
0.1887 | 40.0 | 2680 | 0.1698 | 0.0923 | 0.0281 |
0.1687 | 41.0 | 2747 | 0.1773 | 0.0927 | 0.0288 |
0.1669 | 42.0 | 2814 | 0.1760 | 0.0957 | 0.0291 |
0.1669 | 43.0 | 2881 | 0.1770 | 0.0920 | 0.0288 |
0.143 | 44.0 | 2948 | 0.1734 | 0.0896 | 0.0283 |
0.1555 | 45.0 | 3015 | 0.1748 | 0.0903 | 0.0283 |
0.1555 | 46.0 | 3082 | 0.1756 | 0.0937 | 0.0289 |
0.1543 | 47.0 | 3149 | 0.1714 | 0.0917 | 0.0281 |
0.145 | 48.0 | 3216 | 0.1677 | 0.0927 | 0.0284 |
0.145 | 49.0 | 3283 | 0.1698 | 0.0937 | 0.0287 |
0.1451 | 50.0 | 3350 | 0.1709 | 0.0927 | 0.0290 |
0.1422 | 51.0 | 3417 | 0.1674 | 0.0930 | 0.0286 |
0.1422 | 52.0 | 3484 | 0.1680 | 0.0930 | 0.0281 |
0.1466 | 53.0 | 3551 | 0.1664 | 0.0944 | 0.0281 |
0.1371 | 54.0 | 3618 | 0.1749 | 0.0920 | 0.0282 |
0.1371 | 55.0 | 3685 | 0.1761 | 0.0934 | 0.0285 |
0.1334 | 56.0 | 3752 | 0.1727 | 0.0903 | 0.0280 |
0.1365 | 57.0 | 3819 | 0.1739 | 0.0923 | 0.0284 |
0.1365 | 58.0 | 3886 | 0.1754 | 0.0920 | 0.0279 |
0.1328 | 59.0 | 3953 | 0.1748 | 0.0920 | 0.0285 |
0.1378 | 60.0 | 4020 | 0.1790 | 0.0903 | 0.0280 |
0.1378 | 61.0 | 4087 | 0.1739 | 0.0927 | 0.0280 |
0.1417 | 62.0 | 4154 | 0.1722 | 0.0917 | 0.0283 |
0.13 | 63.0 | 4221 | 0.1718 | 0.0913 | 0.0285 |
0.13 | 64.0 | 4288 | 0.1796 | 0.0940 | 0.0287 |
0.131 | 65.0 | 4355 | 0.1765 | 0.0920 | 0.0289 |
0.131 | 66.0 | 4422 | 0.1740 | 0.0913 | 0.0288 |
0.131 | 67.0 | 4489 | 0.1760 | 0.0968 | 0.0296 |
0.1311 | 68.0 | 4556 | 0.1757 | 0.0913 | 0.0284 |
0.1354 | 69.0 | 4623 | 0.1754 | 0.0920 | 0.0286 |
0.1354 | 70.0 | 4690 | 0.1757 | 0.0920 | 0.0285 |
0.1303 | 71.0 | 4757 | 0.1739 | 0.0882 | 0.0281 |
0.1252 | 72.0 | 4824 | 0.1735 | 0.0893 | 0.0281 |
0.1252 | 73.0 | 4891 | 0.1728 | 0.0913 | 0.0285 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.0.0+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3