<!-- 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-06
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: 0.1446
- Wer: 0.0908
- Cer: 0.0253
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 |
|---|---|---|---|---|---|
| 32.101 | 1.0 | 86 | 3.3100 | 1.0 | 1.0 |
| 8.4627 | 2.0 | 172 | 3.0630 | 1.0 | 1.0 |
| 3.0813 | 3.0 | 258 | 2.9336 | 1.0 | 1.0 |
| 2.9381 | 4.0 | 344 | 2.9115 | 1.0 | 1.0 |
| 2.9064 | 5.0 | 430 | 2.9012 | 1.0 | 1.0 |
| 2.8703 | 6.0 | 516 | 2.5147 | 0.9978 | 0.9959 |
| 2.0805 | 7.0 | 602 | 0.7929 | 0.3573 | 0.0944 |
| 2.0805 | 8.0 | 688 | 0.4395 | 0.2148 | 0.0570 |
| 0.9196 | 9.0 | 774 | 0.3280 | 0.1848 | 0.0505 |
| 0.6289 | 10.0 | 860 | 0.2805 | 0.1708 | 0.0456 |
| 0.4986 | 11.0 | 946 | 0.2635 | 0.1521 | 0.0422 |
| 0.4212 | 12.0 | 1032 | 0.2375 | 0.1378 | 0.0379 |
| 0.3877 | 13.0 | 1118 | 0.2240 | 0.1316 | 0.0364 |
| 0.366 | 14.0 | 1204 | 0.2124 | 0.1216 | 0.0340 |
| 0.366 | 15.0 | 1290 | 0.2000 | 0.1186 | 0.0328 |
| 0.3107 | 16.0 | 1376 | 0.1929 | 0.1102 | 0.0324 |
| 0.3198 | 17.0 | 1462 | 0.1939 | 0.1152 | 0.0326 |
| 0.3339 | 18.0 | 1548 | 0.1875 | 0.1112 | 0.0317 |
| 0.2773 | 19.0 | 1634 | 0.1811 | 0.1063 | 0.0306 |
| 0.2777 | 20.0 | 1720 | 0.1777 | 0.1097 | 0.0304 |
| 0.2412 | 21.0 | 1806 | 0.1709 | 0.1058 | 0.0297 |
| 0.2412 | 22.0 | 1892 | 0.1645 | 0.1014 | 0.0283 |
| 0.2356 | 23.0 | 1978 | 0.1616 | 0.1006 | 0.0282 |
| 0.2426 | 24.0 | 2064 | 0.1623 | 0.0997 | 0.0287 |
| 0.2372 | 25.0 | 2150 | 0.1655 | 0.0999 | 0.0286 |
| 0.2337 | 26.0 | 2236 | 0.1595 | 0.1041 | 0.0290 |
| 0.2221 | 27.0 | 2322 | 0.1560 | 0.0974 | 0.0270 |
| 0.2281 | 28.0 | 2408 | 0.1549 | 0.0955 | 0.0265 |
| 0.2281 | 29.0 | 2494 | 0.1603 | 0.0997 | 0.0277 |
| 0.2196 | 30.0 | 2580 | 0.1553 | 0.0969 | 0.0276 |
| 0.1983 | 31.0 | 2666 | 0.1557 | 0.0960 | 0.0274 |
| 0.2113 | 32.0 | 2752 | 0.1523 | 0.0942 | 0.0266 |
| 0.189 | 33.0 | 2838 | 0.1551 | 0.0972 | 0.0274 |
| 0.1816 | 34.0 | 2924 | 0.1512 | 0.0950 | 0.0261 |
| 0.1823 | 35.0 | 3010 | 0.1561 | 0.0952 | 0.0269 |
| 0.1823 | 36.0 | 3096 | 0.1539 | 0.0965 | 0.0273 |
| 0.1772 | 37.0 | 3182 | 0.1553 | 0.0910 | 0.0263 |
| 0.1712 | 38.0 | 3268 | 0.1540 | 0.0969 | 0.0270 |
| 0.1712 | 39.0 | 3354 | 0.1577 | 0.0974 | 0.0273 |
| 0.1621 | 40.0 | 3440 | 0.1541 | 0.0930 | 0.0262 |
| 0.1883 | 41.0 | 3526 | 0.1545 | 0.0935 | 0.0260 |
| 0.1648 | 42.0 | 3612 | 0.1533 | 0.0942 | 0.0262 |
| 0.1648 | 43.0 | 3698 | 0.1523 | 0.0965 | 0.0266 |
| 0.1911 | 44.0 | 3784 | 0.1537 | 0.0962 | 0.0267 |
| 0.1724 | 45.0 | 3870 | 0.1497 | 0.0928 | 0.0260 |
| 0.156 | 46.0 | 3956 | 0.1492 | 0.0861 | 0.0254 |
| 0.1638 | 47.0 | 4042 | 0.1474 | 0.0906 | 0.0250 |
| 0.16 | 48.0 | 4128 | 0.1476 | 0.0881 | 0.0249 |
| 0.163 | 49.0 | 4214 | 0.1521 | 0.0891 | 0.0252 |
| 0.16 | 50.0 | 4300 | 0.1488 | 0.0913 | 0.0257 |
| 0.16 | 51.0 | 4386 | 0.1494 | 0.0910 | 0.0253 |
| 0.1492 | 52.0 | 4472 | 0.1509 | 0.0903 | 0.0256 |
| 0.1519 | 53.0 | 4558 | 0.1527 | 0.0923 | 0.0255 |
| 0.1471 | 54.0 | 4644 | 0.1468 | 0.0864 | 0.0247 |
| 0.1601 | 55.0 | 4730 | 0.1475 | 0.0898 | 0.0251 |
| 0.1452 | 56.0 | 4816 | 0.1506 | 0.0854 | 0.0243 |
| 0.1391 | 57.0 | 4902 | 0.1477 | 0.0896 | 0.0247 |
| 0.1391 | 58.0 | 4988 | 0.1467 | 0.0891 | 0.0250 |
| 0.1396 | 59.0 | 5074 | 0.1500 | 0.0913 | 0.0255 |
| 0.1424 | 60.0 | 5160 | 0.1474 | 0.0906 | 0.0248 |
| 0.1161 | 61.0 | 5246 | 0.1532 | 0.0864 | 0.0244 |
| 0.1224 | 62.0 | 5332 | 0.1532 | 0.0869 | 0.0247 |
| 0.1303 | 63.0 | 5418 | 0.1500 | 0.0888 | 0.0253 |
| 0.139 | 64.0 | 5504 | 0.1480 | 0.0869 | 0.0246 |
| 0.139 | 65.0 | 5590 | 0.1491 | 0.0898 | 0.0252 |
| 0.1418 | 66.0 | 5676 | 0.1550 | 0.0938 | 0.0256 |
| 0.1228 | 67.0 | 5762 | 0.1477 | 0.0864 | 0.0245 |
| 0.1258 | 68.0 | 5848 | 0.1482 | 0.0859 | 0.0242 |
| 0.1355 | 69.0 | 5934 | 0.1458 | 0.0878 | 0.0248 |
| 0.1297 | 70.0 | 6020 | 0.1446 | 0.0908 | 0.0253 |
| 0.1289 | 71.0 | 6106 | 0.1489 | 0.0878 | 0.0244 |
| 0.1289 | 72.0 | 6192 | 0.1482 | 0.0920 | 0.0257 |
| 0.1241 | 73.0 | 6278 | 0.1494 | 0.0938 | 0.0257 |
| 0.1233 | 74.0 | 6364 | 0.1478 | 0.0883 | 0.0248 |
| 0.1205 | 75.0 | 6450 | 0.1491 | 0.0874 | 0.0246 |
| 0.1302 | 76.0 | 6536 | 0.1534 | 0.0896 | 0.0248 |
| 0.1238 | 77.0 | 6622 | 0.1514 | 0.0896 | 0.0249 |
| 0.1224 | 78.0 | 6708 | 0.1528 | 0.0876 | 0.0247 |
| 0.1224 | 79.0 | 6794 | 0.1531 | 0.0896 | 0.0248 |
| 0.1177 | 80.0 | 6880 | 0.1537 | 0.0888 | 0.0247 |
| 0.1203 | 81.0 | 6966 | 0.1552 | 0.0896 | 0.0246 |
| 0.1095 | 82.0 | 7052 | 0.1552 | 0.0910 | 0.0250 |
| 0.1234 | 83.0 | 7138 | 0.1538 | 0.0869 | 0.0245 |
| 0.1148 | 84.0 | 7224 | 0.1543 | 0.0859 | 0.0244 |
| 0.122 | 85.0 | 7310 | 0.1545 | 0.0881 | 0.0247 |
| 0.122 | 86.0 | 7396 | 0.1536 | 0.0898 | 0.0248 |
| 0.1201 | 87.0 | 7482 | 0.1527 | 0.0920 | 0.0251 |
| 0.1152 | 88.0 | 7568 | 0.1532 | 0.0903 | 0.0251 |
| 0.1178 | 89.0 | 7654 | 0.1517 | 0.0861 | 0.0244 |
| 0.1164 | 90.0 | 7740 | 0.1541 | 0.0901 | 0.0255 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
