<!-- 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-clean-grade-2-4-5
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: 0.1772
- Wer: 0.1024
- Cer: 0.0275
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 |
---|---|---|---|---|---|
20.5968 | 1.0 | 52 | 3.6243 | 1.0 | 1.0 |
7.6648 | 2.0 | 104 | 3.0693 | 1.0 | 1.0 |
7.6648 | 3.0 | 156 | 2.9348 | 1.0 | 1.0 |
3.0453 | 4.0 | 208 | 2.8993 | 1.0 | 1.0 |
3.0453 | 5.0 | 260 | 2.8889 | 1.0 | 1.0 |
2.9153 | 6.0 | 312 | 2.8305 | 1.0 | 1.0 |
2.9153 | 7.0 | 364 | 2.3679 | 1.0 | 0.8415 |
2.6053 | 8.0 | 416 | 1.1410 | 0.8837 | 0.2159 |
2.6053 | 9.0 | 468 | 0.6405 | 0.3048 | 0.0768 |
1.2587 | 10.0 | 520 | 0.4809 | 0.2413 | 0.0608 |
1.2587 | 11.0 | 572 | 0.4091 | 0.2053 | 0.0530 |
0.7468 | 12.0 | 624 | 0.3557 | 0.1904 | 0.0486 |
0.7468 | 13.0 | 676 | 0.3265 | 0.1837 | 0.0465 |
0.549 | 14.0 | 728 | 0.2963 | 0.1740 | 0.0437 |
0.549 | 15.0 | 780 | 0.2850 | 0.1736 | 0.0442 |
0.4682 | 16.0 | 832 | 0.2723 | 0.1606 | 0.0420 |
0.4682 | 17.0 | 884 | 0.2618 | 0.1534 | 0.0402 |
0.4151 | 18.0 | 936 | 0.2405 | 0.1428 | 0.0363 |
0.4151 | 19.0 | 988 | 0.2399 | 0.1413 | 0.0375 |
0.3722 | 20.0 | 1040 | 0.2326 | 0.1399 | 0.0371 |
0.3722 | 21.0 | 1092 | 0.2324 | 0.1332 | 0.0363 |
0.3143 | 22.0 | 1144 | 0.2206 | 0.1298 | 0.0338 |
0.3143 | 23.0 | 1196 | 0.2212 | 0.1202 | 0.0331 |
0.3536 | 24.0 | 1248 | 0.2097 | 0.1183 | 0.0316 |
0.3097 | 25.0 | 1300 | 0.2076 | 0.1197 | 0.0329 |
0.3097 | 26.0 | 1352 | 0.2045 | 0.1154 | 0.0314 |
0.2746 | 27.0 | 1404 | 0.2061 | 0.1178 | 0.0322 |
0.2746 | 28.0 | 1456 | 0.2005 | 0.1087 | 0.0307 |
0.2677 | 29.0 | 1508 | 0.2001 | 0.1101 | 0.0312 |
0.2677 | 30.0 | 1560 | 0.2044 | 0.1135 | 0.0309 |
0.2391 | 31.0 | 1612 | 0.1933 | 0.1115 | 0.0298 |
0.2391 | 32.0 | 1664 | 0.2025 | 0.1106 | 0.0302 |
0.2379 | 33.0 | 1716 | 0.2025 | 0.1096 | 0.0306 |
0.2379 | 34.0 | 1768 | 0.1994 | 0.1062 | 0.0294 |
0.2177 | 35.0 | 1820 | 0.1941 | 0.1062 | 0.0295 |
0.2177 | 36.0 | 1872 | 0.1940 | 0.1087 | 0.0302 |
0.2126 | 37.0 | 1924 | 0.1864 | 0.1115 | 0.0299 |
0.2126 | 38.0 | 1976 | 0.1959 | 0.1058 | 0.0302 |
0.2167 | 39.0 | 2028 | 0.1888 | 0.1058 | 0.0304 |
0.2167 | 40.0 | 2080 | 0.1943 | 0.1072 | 0.0301 |
0.1927 | 41.0 | 2132 | 0.1970 | 0.1067 | 0.0298 |
0.1927 | 42.0 | 2184 | 0.1899 | 0.1067 | 0.0294 |
0.1987 | 43.0 | 2236 | 0.1876 | 0.1024 | 0.0289 |
0.1987 | 44.0 | 2288 | 0.1852 | 0.1048 | 0.0289 |
0.1917 | 45.0 | 2340 | 0.1853 | 0.1043 | 0.0284 |
0.1917 | 46.0 | 2392 | 0.1829 | 0.1043 | 0.0290 |
0.1891 | 47.0 | 2444 | 0.1826 | 0.1043 | 0.0290 |
0.1891 | 48.0 | 2496 | 0.1783 | 0.1024 | 0.0289 |
0.1806 | 49.0 | 2548 | 0.1857 | 0.1043 | 0.0282 |
0.1754 | 50.0 | 2600 | 0.1866 | 0.1058 | 0.0296 |
0.1754 | 51.0 | 2652 | 0.1924 | 0.1024 | 0.0287 |
0.1785 | 52.0 | 2704 | 0.1923 | 0.1087 | 0.0299 |
0.1785 | 53.0 | 2756 | 0.1865 | 0.1062 | 0.0288 |
0.1819 | 54.0 | 2808 | 0.1873 | 0.1029 | 0.0289 |
0.1819 | 55.0 | 2860 | 0.1842 | 0.1029 | 0.0281 |
0.1763 | 56.0 | 2912 | 0.1803 | 0.1029 | 0.0286 |
0.1763 | 57.0 | 2964 | 0.1843 | 0.1005 | 0.0276 |
0.1494 | 58.0 | 3016 | 0.1813 | 0.1005 | 0.0275 |
0.1494 | 59.0 | 3068 | 0.1807 | 0.1029 | 0.0280 |
0.1564 | 60.0 | 3120 | 0.1806 | 0.1010 | 0.0277 |
0.1564 | 61.0 | 3172 | 0.1772 | 0.1024 | 0.0275 |
0.1681 | 62.0 | 3224 | 0.1826 | 0.1043 | 0.0281 |
0.1681 | 63.0 | 3276 | 0.1782 | 0.1034 | 0.0276 |
0.1651 | 64.0 | 3328 | 0.1778 | 0.1 | 0.0272 |
0.1651 | 65.0 | 3380 | 0.1822 | 0.1043 | 0.0278 |
0.1544 | 66.0 | 3432 | 0.1785 | 0.1058 | 0.0287 |
0.1544 | 67.0 | 3484 | 0.1851 | 0.1034 | 0.0284 |
0.144 | 68.0 | 3536 | 0.1860 | 0.1082 | 0.0287 |
0.144 | 69.0 | 3588 | 0.1832 | 0.1087 | 0.0287 |
0.1414 | 70.0 | 3640 | 0.1842 | 0.1082 | 0.0291 |
0.1414 | 71.0 | 3692 | 0.1792 | 0.1062 | 0.0285 |
0.1456 | 72.0 | 3744 | 0.1779 | 0.1062 | 0.0289 |
0.1456 | 73.0 | 3796 | 0.1801 | 0.1034 | 0.0281 |
0.1343 | 74.0 | 3848 | 0.1801 | 0.1053 | 0.0288 |
0.1489 | 75.0 | 3900 | 0.1857 | 0.1034 | 0.0281 |
0.1489 | 76.0 | 3952 | 0.1808 | 0.1053 | 0.0283 |
0.1346 | 77.0 | 4004 | 0.1793 | 0.1034 | 0.0282 |
0.1346 | 78.0 | 4056 | 0.1826 | 0.1058 | 0.0282 |
0.1346 | 79.0 | 4108 | 0.1782 | 0.1034 | 0.0279 |
0.1346 | 80.0 | 4160 | 0.1774 | 0.1019 | 0.0272 |
0.1326 | 81.0 | 4212 | 0.1790 | 0.1010 | 0.0271 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.0.1+cu117
- Datasets 2.14.4
- Tokenizers 0.13.3