<!-- 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-3-4-5
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: 0.0762
- Wer: 0.0684
- Cer: 0.0194
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 |
---|---|---|---|---|---|
17.9625 | 1.0 | 38 | 3.6988 | 1.0 | 1.0 |
17.9625 | 2.0 | 76 | 3.1005 | 1.0 | 1.0 |
5.5893 | 3.0 | 114 | 2.9548 | 1.0 | 1.0 |
5.5893 | 4.0 | 152 | 2.9045 | 1.0 | 1.0 |
5.5893 | 5.0 | 190 | 2.8783 | 1.0 | 1.0 |
2.9471 | 6.0 | 228 | 2.8823 | 1.0 | 1.0 |
2.9471 | 7.0 | 266 | 2.8524 | 1.0 | 1.0 |
2.8717 | 8.0 | 304 | 2.6942 | 1.0 | 1.0 |
2.8717 | 9.0 | 342 | 2.1259 | 0.9995 | 0.7397 |
2.8717 | 10.0 | 380 | 1.0612 | 0.5515 | 0.1423 |
2.1569 | 11.0 | 418 | 0.5953 | 0.2885 | 0.0705 |
2.1569 | 12.0 | 456 | 0.4181 | 0.1979 | 0.0489 |
2.1569 | 13.0 | 494 | 0.3229 | 0.1627 | 0.0395 |
0.8879 | 14.0 | 532 | 0.2739 | 0.1516 | 0.0380 |
0.8879 | 15.0 | 570 | 0.2342 | 0.1410 | 0.0345 |
0.5631 | 16.0 | 608 | 0.2073 | 0.1341 | 0.0327 |
0.5631 | 17.0 | 646 | 0.1926 | 0.1313 | 0.0331 |
0.5631 | 18.0 | 684 | 0.1748 | 0.1285 | 0.0316 |
0.4583 | 19.0 | 722 | 0.1630 | 0.1174 | 0.0296 |
0.4583 | 20.0 | 760 | 0.1534 | 0.1100 | 0.0280 |
0.4583 | 21.0 | 798 | 0.1439 | 0.1049 | 0.0270 |
0.3852 | 22.0 | 836 | 0.1376 | 0.0999 | 0.0271 |
0.3852 | 23.0 | 874 | 0.1323 | 0.0915 | 0.0244 |
0.323 | 24.0 | 912 | 0.1224 | 0.0934 | 0.0237 |
0.323 | 25.0 | 950 | 0.1194 | 0.0906 | 0.0238 |
0.323 | 26.0 | 988 | 0.1148 | 0.0865 | 0.0219 |
0.2892 | 27.0 | 1026 | 0.1165 | 0.0851 | 0.0221 |
0.2892 | 28.0 | 1064 | 0.1097 | 0.0823 | 0.0221 |
0.2968 | 29.0 | 1102 | 0.1068 | 0.0818 | 0.0219 |
0.2968 | 30.0 | 1140 | 0.1043 | 0.0828 | 0.0209 |
0.2968 | 31.0 | 1178 | 0.0994 | 0.0758 | 0.0201 |
0.2756 | 32.0 | 1216 | 0.0979 | 0.0795 | 0.0202 |
0.2756 | 33.0 | 1254 | 0.0945 | 0.0763 | 0.0196 |
0.2756 | 34.0 | 1292 | 0.0951 | 0.0763 | 0.0198 |
0.2348 | 35.0 | 1330 | 0.0927 | 0.0721 | 0.0187 |
0.2348 | 36.0 | 1368 | 0.0932 | 0.0707 | 0.0192 |
0.2474 | 37.0 | 1406 | 0.0921 | 0.0689 | 0.0185 |
0.2474 | 38.0 | 1444 | 0.0905 | 0.0726 | 0.0195 |
0.2474 | 39.0 | 1482 | 0.0882 | 0.0689 | 0.0185 |
0.2344 | 40.0 | 1520 | 0.0862 | 0.0735 | 0.0192 |
0.2344 | 41.0 | 1558 | 0.0869 | 0.0698 | 0.0190 |
0.2344 | 42.0 | 1596 | 0.0852 | 0.0698 | 0.0189 |
0.2129 | 43.0 | 1634 | 0.0837 | 0.0647 | 0.0179 |
0.2129 | 44.0 | 1672 | 0.0834 | 0.0656 | 0.0178 |
0.1985 | 45.0 | 1710 | 0.0826 | 0.0656 | 0.0184 |
0.1985 | 46.0 | 1748 | 0.0806 | 0.0656 | 0.0180 |
0.1985 | 47.0 | 1786 | 0.0808 | 0.0680 | 0.0180 |
0.1975 | 48.0 | 1824 | 0.0798 | 0.0652 | 0.0179 |
0.1975 | 49.0 | 1862 | 0.0811 | 0.0638 | 0.0181 |
0.2037 | 50.0 | 1900 | 0.0816 | 0.0675 | 0.0180 |
0.2037 | 51.0 | 1938 | 0.0797 | 0.0643 | 0.0185 |
0.2037 | 52.0 | 1976 | 0.0808 | 0.0684 | 0.0186 |
0.1869 | 53.0 | 2014 | 0.0803 | 0.0656 | 0.0188 |
0.1869 | 54.0 | 2052 | 0.0801 | 0.0610 | 0.0173 |
0.1869 | 55.0 | 2090 | 0.0801 | 0.0638 | 0.0175 |
0.1808 | 56.0 | 2128 | 0.0794 | 0.0643 | 0.0181 |
0.1808 | 57.0 | 2166 | 0.0800 | 0.0656 | 0.0185 |
0.1693 | 58.0 | 2204 | 0.0805 | 0.0647 | 0.0180 |
0.1693 | 59.0 | 2242 | 0.0817 | 0.0670 | 0.0184 |
0.1693 | 60.0 | 2280 | 0.0820 | 0.0666 | 0.0192 |
0.1736 | 61.0 | 2318 | 0.0791 | 0.0647 | 0.0184 |
0.1736 | 62.0 | 2356 | 0.0801 | 0.0647 | 0.0181 |
0.1736 | 63.0 | 2394 | 0.0805 | 0.0666 | 0.0186 |
0.1726 | 64.0 | 2432 | 0.0798 | 0.0638 | 0.0182 |
0.1726 | 65.0 | 2470 | 0.0813 | 0.0680 | 0.0192 |
0.1648 | 66.0 | 2508 | 0.0812 | 0.0666 | 0.0190 |
0.1648 | 67.0 | 2546 | 0.0801 | 0.0652 | 0.0186 |
0.1648 | 68.0 | 2584 | 0.0808 | 0.0656 | 0.0183 |
0.1501 | 69.0 | 2622 | 0.0797 | 0.0643 | 0.0184 |
0.1501 | 70.0 | 2660 | 0.0802 | 0.0684 | 0.0190 |
0.1501 | 71.0 | 2698 | 0.0801 | 0.0647 | 0.0184 |
0.1584 | 72.0 | 2736 | 0.0798 | 0.0652 | 0.0181 |
0.1584 | 73.0 | 2774 | 0.0790 | 0.0629 | 0.0183 |
0.1465 | 74.0 | 2812 | 0.0783 | 0.0638 | 0.0185 |
0.1465 | 75.0 | 2850 | 0.0786 | 0.0656 | 0.0184 |
0.1465 | 76.0 | 2888 | 0.0790 | 0.0698 | 0.0196 |
0.1589 | 77.0 | 2926 | 0.0780 | 0.0670 | 0.0190 |
0.1589 | 78.0 | 2964 | 0.0784 | 0.0689 | 0.0192 |
0.1445 | 79.0 | 3002 | 0.0792 | 0.0680 | 0.0195 |
0.1445 | 80.0 | 3040 | 0.0787 | 0.0670 | 0.0194 |
0.1445 | 81.0 | 3078 | 0.0776 | 0.0652 | 0.0188 |
0.1378 | 82.0 | 3116 | 0.0771 | 0.0652 | 0.0192 |
0.1378 | 83.0 | 3154 | 0.0775 | 0.0643 | 0.0190 |
0.1378 | 84.0 | 3192 | 0.0786 | 0.0693 | 0.0196 |
0.1538 | 85.0 | 3230 | 0.0767 | 0.0656 | 0.0192 |
0.1538 | 86.0 | 3268 | 0.0770 | 0.0661 | 0.0190 |
0.1351 | 87.0 | 3306 | 0.0776 | 0.0666 | 0.0189 |
0.1351 | 88.0 | 3344 | 0.0768 | 0.0675 | 0.0188 |
0.1351 | 89.0 | 3382 | 0.0768 | 0.0680 | 0.0195 |
0.1433 | 90.0 | 3420 | 0.0766 | 0.0675 | 0.0192 |
0.1433 | 91.0 | 3458 | 0.0770 | 0.0680 | 0.0193 |
0.1433 | 92.0 | 3496 | 0.0778 | 0.0698 | 0.0198 |
0.1368 | 93.0 | 3534 | 0.0767 | 0.0693 | 0.0197 |
0.1368 | 94.0 | 3572 | 0.0775 | 0.0698 | 0.0198 |
0.1338 | 95.0 | 3610 | 0.0769 | 0.0680 | 0.0195 |
0.1338 | 96.0 | 3648 | 0.0764 | 0.0680 | 0.0193 |
0.1338 | 97.0 | 3686 | 0.0763 | 0.0680 | 0.0192 |
0.1384 | 98.0 | 3724 | 0.0762 | 0.0684 | 0.0194 |
0.1384 | 99.0 | 3762 | 0.0765 | 0.0693 | 0.0193 |
0.1341 | 100.0 | 3800 | 0.0766 | 0.0703 | 0.0197 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.0.1+cu117
- Datasets 2.14.4
- Tokenizers 0.13.3