--- license: cc0-1.0 tags: - generated_from_trainer base_model: KBLab/wav2vec2-large-voxrex model-index: - name: wav2vec2-large-voxrex-npsc-nst-bokmaal-fixed results: [] --- # wav2vec2-large-voxrex-npsc-nst-bokmaal-fixed This model is a fine-tuned version of [KBLab/wav2vec2-large-voxrex](https://huggingface.co/KBLab/wav2vec2-large-voxrex) on the None dataset. It achieves the following results on the evaluation set: - Loss: 0.0482 - Wer: 0.0493 ## 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: 0.0001 - train_batch_size: 16 - eval_batch_size: 16 - 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 - lr_scheduler_warmup_steps: 2000 - num_epochs: 15.0 - mixed_precision_training: Native AMP ### Training results | Training Loss | Epoch | Step | Validation Loss | Wer | |:-------------:|:-----:|:------:|:---------------:|:------:| | 6.2179 | 0.05 | 500 | 4.1309 | 0.9998 | | 2.9345 | 0.11 | 1000 | 2.9238 | 1.0 | | 0.7252 | 0.16 | 1500 | 0.3986 | 0.3914 | | 0.5085 | 0.21 | 2000 | 0.2691 | 0.2844 | | 0.397 | 0.26 | 2500 | 0.2210 | 0.2347 | | 0.3632 | 0.32 | 3000 | 0.1961 | 0.2042 | | 0.3393 | 0.37 | 3500 | 0.1793 | 0.1848 | | 0.32 | 0.42 | 4000 | 0.1637 | 0.1682 | | 0.3074 | 0.48 | 4500 | 0.1520 | 0.1592 | | 0.2941 | 0.53 | 5000 | 0.1435 | 0.1532 | | 0.2706 | 0.58 | 5500 | 0.1391 | 0.1468 | | 0.2704 | 0.64 | 6000 | 0.1340 | 0.1385 | | 0.2526 | 0.69 | 6500 | 0.1274 | 0.1336 | | 0.2508 | 0.74 | 7000 | 0.1191 | 0.1290 | | 0.2372 | 0.79 | 7500 | 0.1215 | 0.1252 | | 0.2416 | 0.85 | 8000 | 0.1177 | 0.1246 | | 0.2269 | 0.9 | 8500 | 0.1109 | 0.1191 | | 0.2373 | 0.95 | 9000 | 0.1060 | 0.1202 | | 0.2355 | 1.01 | 9500 | 0.1047 | 0.1186 | | 0.2112 | 1.06 | 10000 | 0.1081 | 0.1123 | | 0.2099 | 1.11 | 10500 | 0.1089 | 0.1109 | | 0.1878 | 1.16 | 11000 | 0.1009 | 0.1102 | | 0.2194 | 1.22 | 11500 | 0.0964 | 0.1121 | | 0.204 | 1.27 | 12000 | 0.0953 | 0.1051 | | 0.184 | 1.32 | 12500 | 0.0934 | 0.1051 | | 0.1896 | 1.38 | 13000 | 0.0968 | 0.1037 | | 0.1849 | 1.43 | 13500 | 0.0917 | 0.1038 | | 0.1829 | 1.48 | 14000 | 0.0907 | 0.1002 | | 0.1781 | 1.53 | 14500 | 0.0898 | 0.0997 | | 0.1838 | 1.59 | 15000 | 0.0873 | 0.1002 | | 0.1868 | 1.64 | 15500 | 0.0883 | 0.0938 | | 0.1888 | 1.69 | 16000 | 0.0850 | 0.0928 | | 0.1767 | 1.75 | 16500 | 0.0851 | 0.0915 | | 0.1759 | 1.8 | 17000 | 0.0864 | 0.0932 | | 0.1778 | 1.85 | 17500 | 0.0842 | 0.0913 | | 0.1612 | 1.91 | 18000 | 0.0831 | 0.0873 | | 0.163 | 1.96 | 18500 | 0.0797 | 0.0940 | | 0.1618 | 2.01 | 19000 | 0.0798 | 0.0902 | | 0.1576 | 2.06 | 19500 | 0.0818 | 0.0883 | | 0.1585 | 2.12 | 20000 | 0.0808 | 0.0879 | | 0.1519 | 2.17 | 20500 | 0.0804 | 0.0868 | | 0.1504 | 2.22 | 21000 | 0.0790 | 0.0834 | | 0.1518 | 2.28 | 21500 | 0.0772 | 0.0832 | | 0.1521 | 2.33 | 22000 | 0.0757 | 0.0820 | | 0.1459 | 2.38 | 22500 | 0.0778 | 0.0815 | | 0.1436 | 2.43 | 23000 | 0.0777 | 0.0808 | | 0.1475 | 2.49 | 23500 | 0.0736 | 0.0800 | | 0.1448 | 2.54 | 24000 | 0.0758 | 0.0809 | | 0.1602 | 2.59 | 24500 | 0.0721 | 0.0802 | | 0.1417 | 2.65 | 25000 | 0.0716 | 0.0795 | | 0.145 | 2.7 | 25500 | 0.0719 | 0.0791 | | 0.1456 | 2.75 | 26000 | 0.0721 | 0.0813 | | 0.1391 | 2.8 | 26500 | 0.0687 | 0.0780 | | 0.1431 | 2.86 | 27000 | 0.0727 | 0.0770 | | 0.24 | 2.91 | 27500 | 0.0699 | 0.0769 | | 0.1333 | 2.96 | 28000 | 0.0701 | 0.0770 | | 0.1305 | 3.02 | 28500 | 0.0715 | 0.0756 | | 0.1366 | 3.07 | 29000 | 0.0700 | 0.0764 | | 0.1354 | 3.12 | 29500 | 0.0674 | 0.0743 | | 0.1269 | 3.18 | 30000 | 0.0689 | 0.0765 | | 0.1259 | 3.23 | 30500 | 0.0703 | 0.0773 | | 0.1246 | 3.28 | 31000 | 0.0672 | 0.0740 | | 0.1281 | 3.33 | 31500 | 0.0683 | 0.0732 | | 0.1356 | 3.39 | 32000 | 0.0686 | 0.0724 | | 0.1258 | 3.44 | 32500 | 0.0663 | 0.0718 | | 0.1305 | 3.49 | 33000 | 0.0680 | 0.0715 | | 0.1274 | 3.55 | 33500 | 0.0682 | 0.0704 | | 0.2169 | 3.6 | 34000 | 0.0663 | 0.0716 | | 0.1301 | 3.65 | 34500 | 0.0654 | 0.0707 | | 0.1242 | 3.7 | 35000 | 0.0653 | 0.0702 | | 0.1284 | 3.76 | 35500 | 0.0641 | 0.0720 | | 0.1204 | 3.81 | 36000 | 0.0642 | 0.0731 | | 0.1256 | 3.86 | 36500 | 0.0645 | 0.0772 | | 0.1147 | 3.92 | 37000 | 0.0659 | 0.0734 | | 0.112 | 3.97 | 37500 | 0.0669 | 0.0728 | | 0.1153 | 4.02 | 38000 | 0.0647 | 0.0738 | | 0.1169 | 4.07 | 38500 | 0.0648 | 0.0681 | | 0.1145 | 4.13 | 39000 | 0.0653 | 0.0686 | | 0.108 | 4.18 | 39500 | 0.0643 | 0.0688 | | 0.1142 | 4.23 | 40000 | 0.0637 | 0.0703 | | 0.1126 | 4.29 | 40500 | 0.0637 | 0.0684 | | 0.118 | 4.34 | 41000 | 0.0647 | 0.0669 | | 0.1128 | 4.39 | 41500 | 0.0640 | 0.0680 | | 0.1159 | 4.45 | 42000 | 0.0613 | 0.0672 | | 0.1149 | 4.5 | 42500 | 0.0617 | 0.0660 | | 0.1111 | 4.55 | 43000 | 0.0606 | 0.0664 | | 0.1162 | 4.6 | 43500 | 0.0605 | 0.0655 | | 0.1087 | 4.66 | 44000 | 0.0614 | 0.0672 | | 0.1161 | 4.71 | 44500 | 0.0618 | 0.0674 | | 0.1093 | 4.76 | 45000 | 0.0617 | 0.0676 | | 0.1157 | 4.82 | 45500 | 0.0619 | 0.0653 | | 0.1077 | 4.87 | 46000 | 0.0606 | 0.0654 | | 0.1059 | 4.92 | 46500 | 0.0597 | 0.0633 | | 0.1031 | 4.97 | 47000 | 0.0608 | 0.0638 | | 0.109 | 5.03 | 47500 | 0.0595 | 0.0632 | | 0.1009 | 5.08 | 48000 | 0.0597 | 0.0658 | | 0.1008 | 5.13 | 48500 | 0.0602 | 0.0674 | | 0.0952 | 5.19 | 49000 | 0.0604 | 0.0677 | | 0.1084 | 5.24 | 49500 | 0.0606 | 0.0672 | | 0.1061 | 5.29 | 50000 | 0.0579 | 0.0654 | | 0.0968 | 5.35 | 50500 | 0.0607 | 0.0666 | | 0.1025 | 5.4 | 51000 | 0.0594 | 0.0655 | | 0.1086 | 5.45 | 51500 | 0.0597 | 0.0677 | | 0.1007 | 5.5 | 52000 | 0.0590 | 0.0632 | | 0.0996 | 5.56 | 52500 | 0.0592 | 0.0633 | | 0.1041 | 5.61 | 53000 | 0.0593 | 0.0637 | | 0.1071 | 5.66 | 53500 | 0.0576 | 0.0627 | | 0.1073 | 5.72 | 54000 | 0.0591 | 0.0637 | | 0.1086 | 5.77 | 54500 | 0.0581 | 0.0619 | | 0.1043 | 5.82 | 55000 | 0.0583 | 0.0613 | | 0.1037 | 5.87 | 55500 | 0.0589 | 0.0614 | | 0.1008 | 5.93 | 56000 | 0.0576 | 0.0604 | | 0.1074 | 5.98 | 56500 | 0.0561 | 0.0608 | | 0.0978 | 6.03 | 57000 | 0.0576 | 0.0610 | | 0.094 | 6.09 | 57500 | 0.0587 | 0.0598 | | 0.0918 | 6.14 | 58000 | 0.0587 | 0.0603 | | 0.0998 | 6.19 | 58500 | 0.0564 | 0.0596 | | 0.1049 | 6.24 | 59000 | 0.0560 | 0.0590 | | 0.0986 | 6.3 | 59500 | 0.0564 | 0.0609 | | 0.092 | 6.35 | 60000 | 0.0565 | 0.0601 | | 0.0944 | 6.4 | 60500 | 0.0552 | 0.0596 | | 0.0908 | 6.46 | 61000 | 0.0567 | 0.0593 | | 0.096 | 6.51 | 61500 | 0.0562 | 0.0588 | | 0.0977 | 6.56 | 62000 | 0.0566 | 0.0601 | | 0.0987 | 6.62 | 62500 | 0.0555 | 0.0594 | | 0.0941 | 6.67 | 63000 | 0.0557 | 0.0603 | | 0.0992 | 6.72 | 63500 | 0.0551 | 0.0603 | | 0.0888 | 6.77 | 64000 | 0.0565 | 0.0623 | | 0.1065 | 6.83 | 64500 | 0.0538 | 0.0598 | | 0.0996 | 6.88 | 65000 | 0.0559 | 0.0601 | | 0.0945 | 6.93 | 65500 | 0.0543 | 0.0608 | | 0.0898 | 6.99 | 66000 | 0.0551 | 0.0600 | | 0.0911 | 7.04 | 66500 | 0.0568 | 0.0598 | | 0.0871 | 7.09 | 67000 | 0.0554 | 0.0605 | | 0.0898 | 7.14 | 67500 | 0.0561 | 0.0600 | | 0.0905 | 7.2 | 68000 | 0.0560 | 0.0582 | | 0.0866 | 7.25 | 68500 | 0.0549 | 0.0614 | | 0.0857 | 7.3 | 69000 | 0.0558 | 0.0594 | | 0.0827 | 7.36 | 69500 | 0.0570 | 0.0603 | | 0.0913 | 7.41 | 70000 | 0.0545 | 0.0592 | | 0.0862 | 7.46 | 70500 | 0.0557 | 0.0591 | | 0.0904 | 7.51 | 71000 | 0.0539 | 0.0575 | | 0.0876 | 7.57 | 71500 | 0.0542 | 0.0587 | | 0.0873 | 7.62 | 72000 | 0.0555 | 0.0576 | | 0.0895 | 7.67 | 72500 | 0.0541 | 0.0586 | | 0.0892 | 7.73 | 73000 | 0.0527 | 0.0576 | | 0.0878 | 7.78 | 73500 | 0.0542 | 0.0588 | | 0.0904 | 7.83 | 74000 | 0.0524 | 0.0577 | | 0.0888 | 7.89 | 74500 | 0.0522 | 0.0582 | | 0.0848 | 7.94 | 75000 | 0.0526 | 0.0569 | | 0.0879 | 7.99 | 75500 | 0.0524 | 0.0584 | | 0.0789 | 8.04 | 76000 | 0.0533 | 0.0596 | | 0.0798 | 8.1 | 76500 | 0.0540 | 0.0592 | | 0.0901 | 8.15 | 77000 | 0.0516 | 0.0590 | | 0.0798 | 8.2 | 77500 | 0.0525 | 0.0571 | | 0.0844 | 8.26 | 78000 | 0.0524 | 0.0567 | | 0.0824 | 8.31 | 78500 | 0.0531 | 0.0560 | | 0.0825 | 8.36 | 79000 | 0.0527 | 0.0558 | | 0.0873 | 8.41 | 79500 | 0.0525 | 0.0564 | | 0.0842 | 8.47 | 80000 | 0.0528 | 0.0557 | | 0.0802 | 8.52 | 80500 | 0.0523 | 0.0559 | | 0.0866 | 8.57 | 81000 | 0.0529 | 0.0562 | | 0.0848 | 8.63 | 81500 | 0.0518 | 0.0567 | | 0.0819 | 8.68 | 82000 | 0.0514 | 0.0560 | | 0.0882 | 8.73 | 82500 | 0.0516 | 0.0564 | | 0.0854 | 8.78 | 83000 | 0.0512 | 0.0555 | | 0.0733 | 8.84 | 83500 | 0.0532 | 0.0558 | | 0.0835 | 8.89 | 84000 | 0.0509 | 0.0552 | | 0.0787 | 8.94 | 84500 | 0.0515 | 0.0547 | | 0.0803 | 9.0 | 85000 | 0.0510 | 0.0563 | | 0.0798 | 9.05 | 85500 | 0.0522 | 0.0558 | | 0.0801 | 9.1 | 86000 | 0.0520 | 0.0586 | | 0.075 | 9.16 | 86500 | 0.0514 | 0.0567 | | 0.0764 | 9.21 | 87000 | 0.0522 | 0.0576 | | 0.0774 | 9.26 | 87500 | 0.0510 | 0.0558 | | 0.1719 | 9.31 | 88000 | 0.0481 | 0.0594 | | 0.0839 | 9.37 | 88500 | 0.0508 | 0.0577 | | 0.0777 | 9.42 | 89000 | 0.0513 | 0.0575 | | 0.0772 | 9.47 | 89500 | 0.0520 | 0.0569 | | 0.0796 | 9.53 | 90000 | 0.0509 | 0.0551 | | 0.0808 | 9.58 | 90500 | 0.0507 | 0.0555 | | 0.08 | 9.63 | 91000 | 0.0501 | 0.0559 | | 0.0757 | 9.68 | 91500 | 0.0498 | 0.0554 | | 0.0724 | 9.74 | 92000 | 0.0505 | 0.0540 | | 0.0782 | 9.79 | 92500 | 0.0507 | 0.0551 | | 0.0839 | 9.84 | 93000 | 0.0523 | 0.0557 | | 0.0779 | 9.9 | 93500 | 0.0500 | 0.0547 | | 0.0768 | 9.95 | 94000 | 0.0509 | 0.0558 | | 0.0786 | 10.0 | 94500 | 0.0506 | 0.0538 | | 0.0719 | 10.06 | 95000 | 0.0492 | 0.0563 | | 0.0721 | 10.11 | 95500 | 0.0501 | 0.0532 | | 0.0725 | 10.16 | 96000 | 0.0500 | 0.0541 | | 0.0724 | 10.21 | 96500 | 0.0506 | 0.0526 | | 0.0677 | 10.27 | 97000 | 0.0503 | 0.0525 | | 0.0726 | 10.32 | 97500 | 0.0499 | 0.0529 | | 0.0779 | 10.37 | 98000 | 0.0500 | 0.0529 | | 0.1584 | 10.43 | 98500 | 0.0509 | 0.0544 | | 0.0781 | 10.48 | 99000 | 0.0502 | 0.0530 | | 0.0677 | 10.53 | 99500 | 0.0503 | 0.0535 | | 0.0729 | 10.58 | 100000 | 0.0498 | 0.0535 | | 0.0741 | 10.64 | 100500 | 0.0493 | 0.0540 | | 0.0698 | 10.69 | 101000 | 0.0501 | 0.0532 | | 0.0711 | 10.74 | 101500 | 0.0485 | 0.0538 | | 0.0763 | 10.8 | 102000 | 0.0500 | 0.0544 | | 0.0745 | 10.85 | 102500 | 0.0491 | 0.0538 | | 0.0749 | 10.9 | 103000 | 0.0488 | 0.0547 | | 0.0746 | 10.95 | 103500 | 0.0500 | 0.0535 | | 0.081 | 11.01 | 104000 | 0.0490 | 0.0535 | | 0.068 | 11.06 | 104500 | 0.0497 | 0.0522 | | 0.07 | 11.11 | 105000 | 0.0499 | 0.0542 | | 0.0706 | 11.17 | 105500 | 0.0487 | 0.0539 | | 0.071 | 11.22 | 106000 | 0.0489 | 0.0535 | | 0.0761 | 11.27 | 106500 | 0.0490 | 0.0531 | | 0.0756 | 11.33 | 107000 | 0.0492 | 0.0536 | | 0.0679 | 11.38 | 107500 | 0.0499 | 0.0530 | | 0.0701 | 11.43 | 108000 | 0.0489 | 0.0523 | | 0.0746 | 11.48 | 108500 | 0.0493 | 0.0526 | | 0.0716 | 11.54 | 109000 | 0.0495 | 0.0529 | | 0.066 | 11.59 | 109500 | 0.0491 | 0.0526 | | 0.0713 | 11.64 | 110000 | 0.0490 | 0.0514 | | 0.0659 | 11.7 | 110500 | 0.0492 | 0.0516 | | 0.0737 | 11.75 | 111000 | 0.0483 | 0.0503 | | 0.0737 | 11.8 | 111500 | 0.0484 | 0.0506 | | 0.0658 | 11.85 | 112000 | 0.0489 | 0.0514 | | 0.0726 | 11.91 | 112500 | 0.0477 | 0.0507 | | 0.0737 | 11.96 | 113000 | 0.0489 | 0.0508 | | 0.0677 | 12.01 | 113500 | 0.0491 | 0.0510 | | 0.0696 | 12.07 | 114000 | 0.0485 | 0.0508 | | 0.0611 | 12.12 | 114500 | 0.0501 | 0.0499 | | 0.0629 | 12.17 | 115000 | 0.0492 | 0.0503 | | 0.0694 | 12.22 | 115500 | 0.0495 | 0.0497 | | 0.1572 | 12.28 | 116000 | 0.0496 | 0.0500 | | 0.0662 | 12.33 | 116500 | 0.0491 | 0.0501 | | 0.0667 | 12.38 | 117000 | 0.0490 | 0.0497 | | 0.0717 | 12.44 | 117500 | 0.0487 | 0.0495 | | 0.0632 | 12.49 | 118000 | 0.0489 | 0.0494 | | 0.0664 | 12.54 | 118500 | 0.0489 | 0.0497 | | 0.0671 | 12.6 | 119000 | 0.0484 | 0.0494 | | 0.0612 | 12.65 | 119500 | 0.0491 | 0.0495 | | 0.0626 | 12.7 | 120000 | 0.0496 | 0.0494 | | 0.0602 | 12.75 | 120500 | 0.0489 | 0.0489 | | 0.0722 | 12.81 | 121000 | 0.0481 | 0.0493 | | 0.0677 | 12.86 | 121500 | 0.0488 | 0.0497 | | 0.0642 | 12.91 | 122000 | 0.0488 | 0.0500 | | 0.0635 | 12.97 | 122500 | 0.0482 | 0.0498 | | 0.0702 | 13.02 | 123000 | 0.0480 | 0.0497 | | 0.0622 | 13.07 | 123500 | 0.0489 | 0.0493 | | 0.0654 | 13.12 | 124000 | 0.0486 | 0.0495 | | 0.0682 | 13.18 | 124500 | 0.0483 | 0.0492 | | 0.062 | 13.23 | 125000 | 0.0486 | 0.0491 | | 0.0666 | 13.28 | 125500 | 0.0490 | 0.0492 | | 0.1656 | 13.34 | 126000 | 0.0487 | 0.0496 | | 0.0633 | 13.39 | 126500 | 0.0487 | 0.0497 | | 0.0578 | 13.44 | 127000 | 0.0488 | 0.0491 | | 0.0595 | 13.49 | 127500 | 0.0487 | 0.0500 | | 0.0645 | 13.55 | 128000 | 0.0482 | 0.0493 | | 0.0722 | 13.6 | 128500 | 0.0483 | 0.0500 | | 0.0664 | 13.65 | 129000 | 0.0485 | 0.0496 | | 0.0627 | 13.71 | 129500 | 0.0486 | 0.0498 | | 0.0605 | 13.76 | 130000 | 0.0486 | 0.0496 | | 0.0678 | 13.81 | 130500 | 0.0481 | 0.0498 | | 0.0695 | 13.87 | 131000 | 0.0486 | 0.0495 | | 0.0609 | 13.92 | 131500 | 0.0477 | 0.0497 | | 0.064 | 13.97 | 132000 | 0.0481 | 0.0496 | | 0.0583 | 14.02 | 132500 | 0.0483 | 0.0499 | | 0.0639 | 14.08 | 133000 | 0.0483 | 0.0499 | | 0.0683 | 14.13 | 133500 | 0.0477 | 0.0495 | | 0.0623 | 14.18 | 134000 | 0.0485 | 0.0492 | | 0.0669 | 14.24 | 134500 | 0.0483 | 0.0492 | | 0.0603 | 14.29 | 135000 | 0.0484 | 0.0494 | | 0.0696 | 14.34 | 135500 | 0.0480 | 0.0494 | | 0.0631 | 14.39 | 136000 | 0.0482 | 0.0494 | | 0.0587 | 14.45 | 136500 | 0.0481 | 0.0493 | | 0.0671 | 14.5 | 137000 | 0.0483 | 0.0492 | | 0.0592 | 14.55 | 137500 | 0.0483 | 0.0493 | | 0.0592 | 14.61 | 138000 | 0.0489 | 0.0494 | | 0.0584 | 14.66 | 138500 | 0.0485 | 0.0495 | | 0.0575 | 14.71 | 139000 | 0.0483 | 0.0495 | | 0.0724 | 14.76 | 139500 | 0.0482 | 0.0494 | | 0.0629 | 14.82 | 140000 | 0.0483 | 0.0493 | | 0.0609 | 14.87 | 140500 | 0.0483 | 0.0493 | | 0.0573 | 14.92 | 141000 | 0.0481 | 0.0494 | | 0.0649 | 14.98 | 141500 | 0.0482 | 0.0493 | ### Framework versions - Transformers 4.18.0 - Pytorch 1.11.0+cu113 - Datasets 2.1.0 - Tokenizers 0.12.1