<!-- 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-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: 0.1319
- Wer: 0.0932
- Cer: 0.0263
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 |
|---|---|---|---|---|---|
| 15.7346 | 0.97 | 16 | 9.2489 | 1.0 | 1.0 |
| 15.7346 | 2.0 | 33 | 4.0322 | 1.0 | 1.0 |
| 15.7346 | 2.97 | 49 | 3.4177 | 1.0 | 1.0 |
| 15.7346 | 4.0 | 66 | 3.1683 | 1.0 | 1.0 |
| 15.7346 | 4.97 | 82 | 3.0383 | 1.0 | 1.0 |
| 15.7346 | 6.0 | 99 | 2.9783 | 1.0 | 1.0 |
| 5.3023 | 6.97 | 115 | 2.9396 | 1.0 | 1.0 |
| 5.3023 | 8.0 | 132 | 2.9159 | 1.0 | 1.0 |
| 5.3023 | 8.97 | 148 | 2.9004 | 1.0 | 1.0 |
| 5.3023 | 10.0 | 165 | 2.8862 | 1.0 | 1.0 |
| 5.3023 | 10.97 | 181 | 2.8701 | 1.0 | 1.0 |
| 5.3023 | 12.0 | 198 | 2.8703 | 1.0 | 1.0 |
| 2.9256 | 12.97 | 214 | 2.8601 | 1.0 | 1.0 |
| 2.9256 | 14.0 | 231 | 2.8912 | 1.0 | 1.0 |
| 2.9256 | 14.97 | 247 | 2.8999 | 1.0 | 1.0 |
| 2.9256 | 16.0 | 264 | 2.8535 | 1.0 | 1.0 |
| 2.9256 | 16.97 | 280 | 2.8630 | 1.0 | 1.0 |
| 2.9256 | 18.0 | 297 | 2.8580 | 1.0 | 1.0 |
| 2.8755 | 18.97 | 313 | 2.8584 | 1.0 | 1.0 |
| 2.8755 | 20.0 | 330 | 2.8444 | 1.0 | 1.0 |
| 2.8755 | 20.97 | 346 | 2.8458 | 1.0 | 1.0 |
| 2.8755 | 22.0 | 363 | 2.8415 | 1.0 | 1.0 |
| 2.8755 | 22.97 | 379 | 2.8386 | 1.0 | 1.0 |
| 2.8755 | 24.0 | 396 | 2.8486 | 1.0 | 1.0 |
| 2.86 | 24.97 | 412 | 2.8298 | 1.0 | 1.0 |
| 2.86 | 26.0 | 429 | 2.8202 | 1.0 | 1.0 |
| 2.86 | 26.97 | 445 | 2.7891 | 1.0 | 1.0 |
| 2.86 | 28.0 | 462 | 2.7456 | 1.0 | 1.0 |
| 2.86 | 28.97 | 478 | 2.6815 | 1.0 | 1.0 |
| 2.86 | 30.0 | 495 | 2.5857 | 1.0 | 1.0 |
| 2.792 | 30.97 | 511 | 2.4622 | 1.0 | 0.9998 |
| 2.792 | 32.0 | 528 | 2.2677 | 0.9989 | 0.8943 |
| 2.792 | 32.97 | 544 | 1.9538 | 1.0 | 0.6955 |
| 2.792 | 34.0 | 561 | 1.5407 | 1.0 | 0.4862 |
| 2.792 | 34.97 | 577 | 1.1644 | 0.8091 | 0.2150 |
| 2.792 | 36.0 | 594 | 0.8820 | 0.5432 | 0.1271 |
| 2.0829 | 36.97 | 610 | 0.6935 | 0.3648 | 0.0828 |
| 2.0829 | 38.0 | 627 | 0.5650 | 0.2614 | 0.0618 |
| 2.0829 | 38.97 | 643 | 0.4996 | 0.2352 | 0.0565 |
| 2.0829 | 40.0 | 660 | 0.4473 | 0.2057 | 0.0513 |
| 2.0829 | 40.97 | 676 | 0.4028 | 0.2023 | 0.0505 |
| 2.0829 | 42.0 | 693 | 0.3718 | 0.1807 | 0.0448 |
| 0.9316 | 42.97 | 709 | 0.3387 | 0.1830 | 0.0446 |
| 0.9316 | 44.0 | 726 | 0.3098 | 0.1761 | 0.0425 |
| 0.9316 | 44.97 | 742 | 0.2873 | 0.1648 | 0.0402 |
| 0.9316 | 46.0 | 759 | 0.2684 | 0.1602 | 0.0382 |
| 0.9316 | 46.97 | 775 | 0.2584 | 0.1523 | 0.0363 |
| 0.9316 | 48.0 | 792 | 0.2493 | 0.15 | 0.0363 |
| 0.5883 | 48.97 | 808 | 0.2373 | 0.1398 | 0.0349 |
| 0.5883 | 50.0 | 825 | 0.2287 | 0.1375 | 0.0347 |
| 0.5883 | 50.97 | 841 | 0.2239 | 0.1386 | 0.0341 |
| 0.5883 | 52.0 | 858 | 0.2143 | 0.1420 | 0.0345 |
| 0.5883 | 52.97 | 874 | 0.2091 | 0.1341 | 0.0337 |
| 0.5883 | 54.0 | 891 | 0.2031 | 0.1307 | 0.0324 |
| 0.4931 | 54.97 | 907 | 0.1941 | 0.1239 | 0.0308 |
| 0.4931 | 56.0 | 924 | 0.1872 | 0.1216 | 0.0306 |
| 0.4931 | 56.97 | 940 | 0.1862 | 0.1227 | 0.0318 |
| 0.4931 | 58.0 | 957 | 0.1850 | 0.1170 | 0.0306 |
| 0.4931 | 58.97 | 973 | 0.1824 | 0.1159 | 0.0320 |
| 0.4931 | 60.0 | 990 | 0.1790 | 0.1193 | 0.0306 |
| 0.4194 | 60.97 | 1006 | 0.1770 | 0.1193 | 0.0306 |
| 0.4194 | 62.0 | 1023 | 0.1735 | 0.1182 | 0.0298 |
| 0.4194 | 62.97 | 1039 | 0.1686 | 0.1170 | 0.0290 |
| 0.4194 | 64.0 | 1056 | 0.1648 | 0.1159 | 0.0294 |
| 0.4194 | 64.97 | 1072 | 0.1617 | 0.1091 | 0.0279 |
| 0.4194 | 66.0 | 1089 | 0.1576 | 0.1068 | 0.0281 |
| 0.3538 | 66.97 | 1105 | 0.1562 | 0.1045 | 0.0277 |
| 0.3538 | 68.0 | 1122 | 0.1548 | 0.1011 | 0.0277 |
| 0.3538 | 68.97 | 1138 | 0.1542 | 0.1045 | 0.0281 |
| 0.3538 | 70.0 | 1155 | 0.1528 | 0.1034 | 0.0285 |
| 0.3538 | 70.97 | 1171 | 0.1500 | 0.1023 | 0.0283 |
| 0.3538 | 72.0 | 1188 | 0.1479 | 0.0932 | 0.0265 |
| 0.313 | 72.97 | 1204 | 0.1461 | 0.0966 | 0.0271 |
| 0.313 | 74.0 | 1221 | 0.1465 | 0.0943 | 0.0267 |
| 0.313 | 74.97 | 1237 | 0.1438 | 0.0932 | 0.0263 |
| 0.313 | 76.0 | 1254 | 0.1433 | 0.0943 | 0.0263 |
| 0.313 | 76.97 | 1270 | 0.1421 | 0.0943 | 0.0253 |
| 0.313 | 78.0 | 1287 | 0.1407 | 0.0909 | 0.0257 |
| 0.3075 | 78.97 | 1303 | 0.1399 | 0.0875 | 0.0251 |
| 0.3075 | 80.0 | 1320 | 0.1378 | 0.0875 | 0.0248 |
| 0.3075 | 80.97 | 1336 | 0.1366 | 0.0886 | 0.0244 |
| 0.3075 | 82.0 | 1353 | 0.1376 | 0.0909 | 0.0257 |
| 0.3075 | 82.97 | 1369 | 0.1344 | 0.0864 | 0.0253 |
| 0.3075 | 84.0 | 1386 | 0.1348 | 0.0943 | 0.0261 |
| 0.2861 | 84.97 | 1402 | 0.1349 | 0.0955 | 0.0261 |
| 0.2861 | 86.0 | 1419 | 0.1339 | 0.0920 | 0.0255 |
| 0.2861 | 86.97 | 1435 | 0.1335 | 0.0943 | 0.0261 |
| 0.2861 | 88.0 | 1452 | 0.1330 | 0.0955 | 0.0259 |
| 0.2861 | 88.97 | 1468 | 0.1326 | 0.0955 | 0.0257 |
| 0.2861 | 90.0 | 1485 | 0.1322 | 0.0955 | 0.0257 |
| 0.2778 | 90.97 | 1501 | 0.1320 | 0.0943 | 0.0259 |
| 0.2778 | 92.0 | 1518 | 0.1329 | 0.0966 | 0.0263 |
| 0.2778 | 92.97 | 1534 | 0.1328 | 0.0955 | 0.0267 |
| 0.2778 | 94.0 | 1551 | 0.1329 | 0.0966 | 0.0265 |
| 0.2778 | 94.97 | 1567 | 0.1324 | 0.0943 | 0.0265 |
| 0.2778 | 96.0 | 1584 | 0.1319 | 0.0932 | 0.0263 |
| 0.3134 | 96.97 | 1600 | 0.1321 | 0.0932 | 0.0265 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.0.1+cu117
- Datasets 2.14.4
- Tokenizers 0.13.3
