<!-- 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-10
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: 0.1658
- Wer: 0.0947
- Cer: 0.0291
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 |
---|---|---|---|---|---|
21.131 | 1.0 | 67 | 3.3444 | 1.0 | 1.0 |
7.1783 | 2.0 | 134 | 2.9805 | 1.0 | 1.0 |
3.0142 | 3.0 | 201 | 2.9060 | 1.0 | 1.0 |
3.0142 | 4.0 | 268 | 2.8839 | 1.0 | 1.0 |
2.9114 | 5.0 | 335 | 2.8077 | 1.0 | 1.0 |
2.7413 | 6.0 | 402 | 1.8757 | 1.0 | 0.5239 |
2.7413 | 7.0 | 469 | 0.6953 | 0.3608 | 0.0901 |
1.3808 | 8.0 | 536 | 0.4579 | 0.2221 | 0.0576 |
0.7472 | 9.0 | 603 | 0.3593 | 0.2 | 0.0523 |
0.7472 | 10.0 | 670 | 0.3176 | 0.1816 | 0.0485 |
0.564 | 11.0 | 737 | 0.2996 | 0.1680 | 0.0467 |
0.47 | 12.0 | 804 | 0.2680 | 0.1564 | 0.0427 |
0.47 | 13.0 | 871 | 0.2549 | 0.1455 | 0.0397 |
0.3958 | 14.0 | 938 | 0.2384 | 0.1322 | 0.0371 |
0.346 | 15.0 | 1005 | 0.2362 | 0.1233 | 0.0361 |
0.346 | 16.0 | 1072 | 0.2118 | 0.1145 | 0.0338 |
0.3286 | 17.0 | 1139 | 0.2110 | 0.1141 | 0.0336 |
0.2927 | 18.0 | 1206 | 0.2048 | 0.1114 | 0.0329 |
0.2927 | 19.0 | 1273 | 0.1992 | 0.1022 | 0.0316 |
0.2996 | 20.0 | 1340 | 0.2000 | 0.1039 | 0.0317 |
0.2532 | 21.0 | 1407 | 0.1981 | 0.1015 | 0.0311 |
0.2532 | 22.0 | 1474 | 0.1961 | 0.1046 | 0.0311 |
0.2465 | 23.0 | 1541 | 0.1902 | 0.1073 | 0.0314 |
0.2369 | 24.0 | 1608 | 0.1871 | 0.0957 | 0.0302 |
0.2369 | 25.0 | 1675 | 0.1850 | 0.0964 | 0.0299 |
0.2224 | 26.0 | 1742 | 0.1817 | 0.0985 | 0.0298 |
0.2352 | 27.0 | 1809 | 0.1824 | 0.0971 | 0.0304 |
0.2352 | 28.0 | 1876 | 0.1859 | 0.0947 | 0.0303 |
0.2191 | 29.0 | 1943 | 0.1822 | 0.0923 | 0.0300 |
0.2064 | 30.0 | 2010 | 0.1773 | 0.0934 | 0.0297 |
0.2064 | 31.0 | 2077 | 0.1794 | 0.0947 | 0.0293 |
0.1964 | 32.0 | 2144 | 0.1763 | 0.0934 | 0.0296 |
0.1962 | 33.0 | 2211 | 0.1748 | 0.0927 | 0.0301 |
0.1962 | 34.0 | 2278 | 0.1789 | 0.0954 | 0.0296 |
0.1766 | 35.0 | 2345 | 0.1729 | 0.0930 | 0.0285 |
0.1782 | 36.0 | 2412 | 0.1738 | 0.0913 | 0.0285 |
0.1782 | 37.0 | 2479 | 0.1714 | 0.0957 | 0.0293 |
0.1835 | 38.0 | 2546 | 0.1739 | 0.0947 | 0.0296 |
0.1907 | 39.0 | 2613 | 0.1725 | 0.0913 | 0.0290 |
0.1907 | 40.0 | 2680 | 0.1718 | 0.0937 | 0.0294 |
0.1667 | 41.0 | 2747 | 0.1718 | 0.0954 | 0.0297 |
0.169 | 42.0 | 2814 | 0.1753 | 0.0927 | 0.0292 |
0.169 | 43.0 | 2881 | 0.1737 | 0.0947 | 0.0295 |
0.1425 | 44.0 | 2948 | 0.1684 | 0.0940 | 0.0292 |
0.1541 | 45.0 | 3015 | 0.1751 | 0.0940 | 0.0293 |
0.1541 | 46.0 | 3082 | 0.1735 | 0.0913 | 0.0286 |
0.1539 | 47.0 | 3149 | 0.1749 | 0.0961 | 0.0292 |
0.1456 | 48.0 | 3216 | 0.1703 | 0.0930 | 0.0290 |
0.1456 | 49.0 | 3283 | 0.1705 | 0.0906 | 0.0285 |
0.1449 | 50.0 | 3350 | 0.1702 | 0.0940 | 0.0289 |
0.1418 | 51.0 | 3417 | 0.1675 | 0.0903 | 0.0280 |
0.1418 | 52.0 | 3484 | 0.1708 | 0.0910 | 0.0285 |
0.1428 | 53.0 | 3551 | 0.1658 | 0.0947 | 0.0291 |
0.1388 | 54.0 | 3618 | 0.1715 | 0.0927 | 0.0287 |
0.1388 | 55.0 | 3685 | 0.1692 | 0.0889 | 0.0278 |
0.1321 | 56.0 | 3752 | 0.1695 | 0.0882 | 0.0275 |
0.135 | 57.0 | 3819 | 0.1715 | 0.0940 | 0.0283 |
0.135 | 58.0 | 3886 | 0.1676 | 0.0903 | 0.0278 |
0.1335 | 59.0 | 3953 | 0.1687 | 0.0930 | 0.0282 |
0.1378 | 60.0 | 4020 | 0.1743 | 0.0923 | 0.0286 |
0.1378 | 61.0 | 4087 | 0.1664 | 0.0920 | 0.0276 |
0.142 | 62.0 | 4154 | 0.1679 | 0.0910 | 0.0274 |
0.1299 | 63.0 | 4221 | 0.1702 | 0.0913 | 0.0278 |
0.1299 | 64.0 | 4288 | 0.1705 | 0.0937 | 0.0283 |
0.1303 | 65.0 | 4355 | 0.1710 | 0.0913 | 0.0281 |
0.1305 | 66.0 | 4422 | 0.1672 | 0.0927 | 0.0280 |
0.1305 | 67.0 | 4489 | 0.1664 | 0.0974 | 0.0291 |
0.1312 | 68.0 | 4556 | 0.1659 | 0.0985 | 0.0287 |
0.1335 | 69.0 | 4623 | 0.1664 | 0.0910 | 0.0275 |
0.1335 | 70.0 | 4690 | 0.1680 | 0.0964 | 0.0286 |
0.1298 | 71.0 | 4757 | 0.1714 | 0.0951 | 0.0285 |
0.1245 | 72.0 | 4824 | 0.1713 | 0.0906 | 0.0275 |
0.1245 | 73.0 | 4891 | 0.1692 | 0.0927 | 0.0280 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3