<!-- 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-3
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: 0.0955
- Wer: 0.0867
- Cer: 0.0224
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 |
---|---|---|---|---|---|
22.3909 | 0.98 | 21 | 8.3600 | 1.0 | 1.0 |
22.3909 | 2.0 | 43 | 3.7086 | 1.0 | 1.0 |
22.3909 | 2.98 | 64 | 3.3101 | 1.0 | 1.0 |
22.3909 | 4.0 | 86 | 3.1299 | 1.0 | 1.0 |
6.6597 | 4.98 | 107 | 3.0237 | 1.0 | 1.0 |
6.6597 | 6.0 | 129 | 2.9785 | 1.0 | 1.0 |
6.6597 | 6.98 | 150 | 2.9219 | 1.0 | 1.0 |
6.6597 | 8.0 | 172 | 2.9057 | 1.0 | 1.0 |
6.6597 | 8.98 | 193 | 2.8967 | 1.0 | 1.0 |
2.9937 | 10.0 | 215 | 2.8868 | 1.0 | 1.0 |
2.9937 | 10.98 | 236 | 2.8936 | 1.0 | 1.0 |
2.9937 | 12.0 | 258 | 2.8946 | 1.0 | 1.0 |
2.9937 | 12.98 | 279 | 2.8602 | 1.0 | 1.0 |
2.8963 | 14.0 | 301 | 2.8328 | 1.0 | 1.0 |
2.8963 | 14.98 | 322 | 2.6860 | 1.0 | 1.0 |
2.8963 | 16.0 | 344 | 2.4361 | 0.9970 | 0.9481 |
2.8963 | 16.98 | 365 | 2.0512 | 1.0 | 0.7348 |
2.8963 | 18.0 | 387 | 1.4315 | 0.9970 | 0.3725 |
2.4322 | 18.98 | 408 | 1.0304 | 0.8684 | 0.2096 |
2.4322 | 20.0 | 430 | 0.7321 | 0.3500 | 0.0873 |
2.4322 | 20.98 | 451 | 0.5688 | 0.2632 | 0.0653 |
2.4322 | 22.0 | 473 | 0.4880 | 0.2233 | 0.0578 |
2.4322 | 22.98 | 494 | 0.4193 | 0.1894 | 0.0469 |
1.0966 | 24.0 | 516 | 0.3687 | 0.1765 | 0.0471 |
1.0966 | 24.98 | 537 | 0.3331 | 0.1685 | 0.0431 |
1.0966 | 26.0 | 559 | 0.3042 | 0.1625 | 0.0402 |
1.0966 | 26.98 | 580 | 0.2854 | 0.1645 | 0.0412 |
0.6624 | 28.0 | 602 | 0.2663 | 0.1595 | 0.0404 |
0.6624 | 28.98 | 623 | 0.2486 | 0.1466 | 0.0360 |
0.6624 | 30.0 | 645 | 0.2348 | 0.1515 | 0.0364 |
0.6624 | 30.98 | 666 | 0.2253 | 0.1486 | 0.0372 |
0.6624 | 32.0 | 688 | 0.2129 | 0.1306 | 0.0330 |
0.4672 | 32.98 | 709 | 0.2088 | 0.1376 | 0.0344 |
0.4672 | 34.0 | 731 | 0.1984 | 0.1256 | 0.0319 |
0.4672 | 34.98 | 752 | 0.1944 | 0.1256 | 0.0315 |
0.4672 | 36.0 | 774 | 0.1923 | 0.1176 | 0.0307 |
0.4672 | 36.98 | 795 | 0.1881 | 0.1107 | 0.0293 |
0.4045 | 38.0 | 817 | 0.1807 | 0.1107 | 0.0289 |
0.4045 | 38.98 | 838 | 0.1716 | 0.1127 | 0.0293 |
0.4045 | 40.0 | 860 | 0.1670 | 0.1167 | 0.0287 |
0.4045 | 40.98 | 881 | 0.1652 | 0.1206 | 0.0295 |
0.3396 | 42.0 | 903 | 0.1618 | 0.1196 | 0.0297 |
0.3396 | 42.98 | 924 | 0.1566 | 0.1077 | 0.0275 |
0.3396 | 44.0 | 946 | 0.1516 | 0.1067 | 0.0273 |
0.3396 | 44.98 | 967 | 0.1490 | 0.1087 | 0.0283 |
0.3396 | 46.0 | 989 | 0.1468 | 0.1077 | 0.0281 |
0.2842 | 46.98 | 1010 | 0.1397 | 0.1057 | 0.0273 |
0.2842 | 48.0 | 1032 | 0.1363 | 0.1057 | 0.0269 |
0.2842 | 48.98 | 1053 | 0.1339 | 0.1027 | 0.0261 |
0.2842 | 50.0 | 1075 | 0.1339 | 0.1077 | 0.0275 |
0.2842 | 50.98 | 1096 | 0.1310 | 0.1017 | 0.0255 |
0.2823 | 52.0 | 1118 | 0.1312 | 0.1057 | 0.0261 |
0.2823 | 52.98 | 1139 | 0.1297 | 0.1027 | 0.0261 |
0.2823 | 54.0 | 1161 | 0.1273 | 0.1017 | 0.0267 |
0.2823 | 54.98 | 1182 | 0.1249 | 0.0997 | 0.0263 |
0.2645 | 56.0 | 1204 | 0.1236 | 0.0977 | 0.0271 |
0.2645 | 56.98 | 1225 | 0.1213 | 0.0997 | 0.0265 |
0.2645 | 58.0 | 1247 | 0.1184 | 0.0977 | 0.0257 |
0.2645 | 58.98 | 1268 | 0.1175 | 0.0957 | 0.0255 |
0.2645 | 60.0 | 1290 | 0.1154 | 0.0977 | 0.0247 |
0.2478 | 60.98 | 1311 | 0.1129 | 0.0987 | 0.0251 |
0.2478 | 62.0 | 1333 | 0.1118 | 0.0987 | 0.0243 |
0.2478 | 62.98 | 1354 | 0.1094 | 0.0957 | 0.0239 |
0.2478 | 64.0 | 1376 | 0.1100 | 0.0967 | 0.0243 |
0.2478 | 64.98 | 1397 | 0.1082 | 0.0907 | 0.0234 |
0.2222 | 66.0 | 1419 | 0.1091 | 0.0967 | 0.0245 |
0.2222 | 66.98 | 1440 | 0.1078 | 0.0977 | 0.0249 |
0.2222 | 68.0 | 1462 | 0.1082 | 0.0967 | 0.0247 |
0.2222 | 68.98 | 1483 | 0.1076 | 0.0977 | 0.0259 |
0.2103 | 70.0 | 1505 | 0.1070 | 0.0977 | 0.0251 |
0.2103 | 70.98 | 1526 | 0.1051 | 0.0907 | 0.0237 |
0.2103 | 72.0 | 1548 | 0.1038 | 0.0867 | 0.0236 |
0.2103 | 72.98 | 1569 | 0.1023 | 0.0897 | 0.0241 |
0.2103 | 74.0 | 1591 | 0.1018 | 0.0907 | 0.0245 |
0.1896 | 74.98 | 1612 | 0.1046 | 0.0907 | 0.0241 |
0.1896 | 76.0 | 1634 | 0.1028 | 0.0887 | 0.0241 |
0.1896 | 76.98 | 1655 | 0.1014 | 0.0867 | 0.0237 |
0.1896 | 78.0 | 1677 | 0.1013 | 0.0857 | 0.0234 |
0.1896 | 78.98 | 1698 | 0.1016 | 0.0877 | 0.0237 |
0.2072 | 80.0 | 1720 | 0.1008 | 0.0937 | 0.0243 |
0.2072 | 80.98 | 1741 | 0.1006 | 0.0917 | 0.0237 |
0.2072 | 82.0 | 1763 | 0.0998 | 0.0927 | 0.0237 |
0.2072 | 82.98 | 1784 | 0.0990 | 0.0927 | 0.0239 |
0.1824 | 84.0 | 1806 | 0.0974 | 0.0937 | 0.0241 |
0.1824 | 84.98 | 1827 | 0.0976 | 0.0897 | 0.0230 |
0.1824 | 86.0 | 1849 | 0.0975 | 0.0907 | 0.0232 |
0.1824 | 86.98 | 1870 | 0.0980 | 0.0927 | 0.0232 |
0.1824 | 88.0 | 1892 | 0.0976 | 0.0927 | 0.0234 |
0.1791 | 88.98 | 1913 | 0.0969 | 0.0917 | 0.0232 |
0.1791 | 90.0 | 1935 | 0.0965 | 0.0907 | 0.0230 |
0.1791 | 90.98 | 1956 | 0.0964 | 0.0907 | 0.0234 |
0.1791 | 92.0 | 1978 | 0.0965 | 0.0907 | 0.0234 |
0.1791 | 92.98 | 1999 | 0.0960 | 0.0877 | 0.0226 |
0.1788 | 94.0 | 2021 | 0.0960 | 0.0887 | 0.0228 |
0.1788 | 94.98 | 2042 | 0.0958 | 0.0877 | 0.0226 |
0.1788 | 96.0 | 2064 | 0.0955 | 0.0867 | 0.0224 |
0.1788 | 96.98 | 2085 | 0.0955 | 0.0857 | 0.0222 |
0.181 | 97.67 | 2100 | 0.0955 | 0.0867 | 0.0224 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.0.1+cu117
- Datasets 2.14.4
- Tokenizers 0.13.3