<!-- 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-grade-2-5
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: 0.1712
- Wer: 0.0876
- Cer: 0.0271
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 |
---|---|---|---|---|---|
35.5693 | 1.0 | 58 | 3.5640 | 1.0 | 1.0 |
8.0772 | 2.0 | 116 | 3.0942 | 1.0 | 1.0 |
8.0772 | 3.0 | 174 | 2.9408 | 1.0 | 1.0 |
3.0422 | 4.0 | 232 | 2.9205 | 1.0 | 1.0 |
3.0422 | 5.0 | 290 | 2.9105 | 1.0 | 1.0 |
2.9326 | 6.0 | 348 | 2.9032 | 1.0 | 1.0 |
2.9098 | 7.0 | 406 | 2.8871 | 1.0 | 1.0 |
2.9098 | 8.0 | 464 | 2.4065 | 1.0 | 0.8613 |
2.575 | 9.0 | 522 | 1.0328 | 0.6076 | 0.1524 |
2.575 | 10.0 | 580 | 0.5981 | 0.2838 | 0.0767 |
1.2138 | 11.0 | 638 | 0.4772 | 0.2248 | 0.0633 |
1.2138 | 12.0 | 696 | 0.4108 | 0.1966 | 0.0570 |
0.73 | 13.0 | 754 | 0.3665 | 0.1806 | 0.0519 |
0.606 | 14.0 | 812 | 0.3326 | 0.1749 | 0.0501 |
0.606 | 15.0 | 870 | 0.3109 | 0.1596 | 0.0457 |
0.5053 | 16.0 | 928 | 0.2984 | 0.1589 | 0.0450 |
0.5053 | 17.0 | 986 | 0.2855 | 0.1478 | 0.0430 |
0.4441 | 18.0 | 1044 | 0.2633 | 0.1383 | 0.0406 |
0.3808 | 19.0 | 1102 | 0.2658 | 0.1295 | 0.0395 |
0.3808 | 20.0 | 1160 | 0.2538 | 0.1257 | 0.0378 |
0.3604 | 21.0 | 1218 | 0.2434 | 0.1204 | 0.0373 |
0.3604 | 22.0 | 1276 | 0.2411 | 0.1150 | 0.0359 |
0.3292 | 23.0 | 1334 | 0.2315 | 0.1109 | 0.0352 |
0.3292 | 24.0 | 1392 | 0.2302 | 0.1093 | 0.0343 |
0.3007 | 25.0 | 1450 | 0.2304 | 0.1063 | 0.0340 |
0.2834 | 26.0 | 1508 | 0.2276 | 0.1059 | 0.0337 |
0.2834 | 27.0 | 1566 | 0.2195 | 0.1074 | 0.0327 |
0.266 | 28.0 | 1624 | 0.2170 | 0.1055 | 0.0321 |
0.266 | 29.0 | 1682 | 0.2174 | 0.1067 | 0.0331 |
0.2483 | 30.0 | 1740 | 0.2122 | 0.1036 | 0.0322 |
0.2483 | 31.0 | 1798 | 0.2132 | 0.1051 | 0.0326 |
0.2517 | 32.0 | 1856 | 0.2152 | 0.1036 | 0.0319 |
0.2646 | 33.0 | 1914 | 0.2061 | 0.1029 | 0.0316 |
0.2646 | 34.0 | 1972 | 0.2048 | 0.1006 | 0.0309 |
0.2365 | 35.0 | 2030 | 0.1972 | 0.1025 | 0.0309 |
0.2365 | 36.0 | 2088 | 0.1983 | 0.0990 | 0.0303 |
0.2005 | 37.0 | 2146 | 0.1903 | 0.0975 | 0.0309 |
0.2185 | 38.0 | 2204 | 0.1951 | 0.096 | 0.0307 |
0.2185 | 39.0 | 2262 | 0.1915 | 0.0933 | 0.0298 |
0.2197 | 40.0 | 2320 | 0.1874 | 0.0956 | 0.0299 |
0.2197 | 41.0 | 2378 | 0.1877 | 0.0952 | 0.0298 |
0.2014 | 42.0 | 2436 | 0.1845 | 0.0971 | 0.0304 |
0.2014 | 43.0 | 2494 | 0.1820 | 0.0933 | 0.0291 |
0.2206 | 44.0 | 2552 | 0.1814 | 0.096 | 0.0298 |
0.202 | 45.0 | 2610 | 0.1786 | 0.0910 | 0.0282 |
0.202 | 46.0 | 2668 | 0.1822 | 0.0956 | 0.0292 |
0.1941 | 47.0 | 2726 | 0.1847 | 0.0964 | 0.0295 |
0.1941 | 48.0 | 2784 | 0.1806 | 0.096 | 0.0300 |
0.1906 | 49.0 | 2842 | 0.1850 | 0.0933 | 0.0285 |
0.1839 | 50.0 | 2900 | 0.1860 | 0.0990 | 0.0294 |
0.1839 | 51.0 | 2958 | 0.1817 | 0.0952 | 0.0292 |
0.1869 | 52.0 | 3016 | 0.1830 | 0.0968 | 0.0298 |
0.1869 | 53.0 | 3074 | 0.1816 | 0.0964 | 0.0295 |
0.1795 | 54.0 | 3132 | 0.1801 | 0.0952 | 0.0287 |
0.1795 | 55.0 | 3190 | 0.1818 | 0.0941 | 0.0291 |
0.1802 | 56.0 | 3248 | 0.1790 | 0.0945 | 0.0290 |
0.1786 | 57.0 | 3306 | 0.1776 | 0.0895 | 0.0279 |
0.1786 | 58.0 | 3364 | 0.1822 | 0.0949 | 0.0290 |
0.1485 | 59.0 | 3422 | 0.1751 | 0.0903 | 0.0282 |
0.1485 | 60.0 | 3480 | 0.1777 | 0.0899 | 0.0284 |
0.1874 | 61.0 | 3538 | 0.1732 | 0.0899 | 0.0282 |
0.1874 | 62.0 | 3596 | 0.1801 | 0.0907 | 0.0279 |
0.1547 | 63.0 | 3654 | 0.1765 | 0.0922 | 0.0290 |
0.1617 | 64.0 | 3712 | 0.1750 | 0.0884 | 0.0278 |
0.1617 | 65.0 | 3770 | 0.1723 | 0.0895 | 0.0281 |
0.1515 | 66.0 | 3828 | 0.1753 | 0.0888 | 0.0279 |
0.1515 | 67.0 | 3886 | 0.1758 | 0.0888 | 0.0282 |
0.1455 | 68.0 | 3944 | 0.1712 | 0.0876 | 0.0271 |
0.1384 | 69.0 | 4002 | 0.1734 | 0.0888 | 0.0273 |
0.1384 | 70.0 | 4060 | 0.1741 | 0.0895 | 0.0274 |
0.1454 | 71.0 | 4118 | 0.1769 | 0.0876 | 0.0279 |
0.1454 | 72.0 | 4176 | 0.1759 | 0.0884 | 0.0278 |
0.1526 | 73.0 | 4234 | 0.1749 | 0.0876 | 0.0276 |
0.1526 | 74.0 | 4292 | 0.1724 | 0.0888 | 0.0275 |
0.1412 | 75.0 | 4350 | 0.1753 | 0.0884 | 0.0278 |
0.1528 | 76.0 | 4408 | 0.1779 | 0.0865 | 0.0274 |
0.1528 | 77.0 | 4466 | 0.1753 | 0.0853 | 0.0276 |
0.1399 | 78.0 | 4524 | 0.1795 | 0.0869 | 0.0275 |
0.1399 | 79.0 | 4582 | 0.1766 | 0.0895 | 0.0285 |
0.1363 | 80.0 | 4640 | 0.1770 | 0.0888 | 0.0282 |
0.1363 | 81.0 | 4698 | 0.1760 | 0.0891 | 0.0278 |
0.1471 | 82.0 | 4756 | 0.1753 | 0.0895 | 0.0280 |
0.1344 | 83.0 | 4814 | 0.1754 | 0.0876 | 0.0276 |
0.1344 | 84.0 | 4872 | 0.1772 | 0.0869 | 0.0279 |
0.142 | 85.0 | 4930 | 0.1792 | 0.0869 | 0.0276 |
0.142 | 86.0 | 4988 | 0.1782 | 0.0865 | 0.0276 |
0.1313 | 87.0 | 5046 | 0.1768 | 0.088 | 0.0276 |
0.1406 | 88.0 | 5104 | 0.1783 | 0.0899 | 0.0282 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.0.1+cu117
- Datasets 2.14.4
- Tokenizers 0.13.3