--- license: cc0-1.0 tags: - generated_from_trainer base_model: KBLab/wav2vec2-large-voxrex model-index: - name: wav2vec2-large-voxrex-npsc-nst results: [] --- # wav2vec2-large-voxrex-npsc-nst 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: 0.0475 - Wer: 0.0514 ## 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 | |:-------------:|:-----:|:------:|:---------------:|:------:| | 3.3888 | 0.05 | 500 | 3.2558 | 1.0 | | 2.7683 | 0.11 | 1000 | 2.4163 | 1.0000 | | 0.6279 | 0.16 | 1500 | 0.3610 | 0.3608 | | 0.5093 | 0.21 | 2000 | 0.2610 | 0.2776 | | 0.4024 | 0.26 | 2500 | 0.2219 | 0.2303 | | 0.3705 | 0.32 | 3000 | 0.1940 | 0.2043 | | 0.3588 | 0.37 | 3500 | 0.1806 | 0.1822 | | 0.3312 | 0.42 | 4000 | 0.1611 | 0.1736 | | 0.3062 | 0.47 | 4500 | 0.1571 | 0.1619 | | 0.2838 | 0.53 | 5000 | 0.1482 | 0.1552 | | 0.2896 | 0.58 | 5500 | 0.1406 | 0.1482 | | 0.2704 | 0.63 | 6000 | 0.1311 | 0.1467 | | 0.263 | 0.69 | 6500 | 0.1258 | 0.1406 | | 0.2574 | 0.74 | 7000 | 0.1252 | 0.1343 | | 0.252 | 0.79 | 7500 | 0.1162 | 0.1279 | | 0.2355 | 0.84 | 8000 | 0.1161 | 0.1275 | | 0.2381 | 0.9 | 8500 | 0.1095 | 0.1247 | | 0.2354 | 0.95 | 9000 | 0.1106 | 0.1250 | | 0.234 | 1.0 | 9500 | 0.1044 | 0.1186 | | 0.2094 | 1.05 | 10000 | 0.1052 | 0.1157 | | 0.2088 | 1.11 | 10500 | 0.1026 | 0.1158 | | 0.2123 | 1.16 | 11000 | 0.0998 | 0.1120 | | 0.3087 | 1.21 | 11500 | 0.0971 | 0.1108 | | 0.1995 | 1.26 | 12000 | 0.0973 | 0.1085 | | 0.1989 | 1.32 | 12500 | 0.0928 | 0.1063 | | 0.1993 | 1.37 | 13000 | 0.0920 | 0.1064 | | 0.1996 | 1.42 | 13500 | 0.0904 | 0.1050 | | 0.1917 | 1.48 | 14000 | 0.0895 | 0.1051 | | 0.1857 | 1.53 | 14500 | 0.0889 | 0.1038 | | 0.1871 | 1.58 | 15000 | 0.0867 | 0.1054 | | 0.2047 | 1.63 | 15500 | 0.0866 | 0.1017 | | 0.1845 | 1.69 | 16000 | 0.0865 | 0.1007 | | 0.178 | 1.74 | 16500 | 0.0835 | 0.0999 | | 0.1741 | 1.79 | 17000 | 0.0838 | 0.0985 | | 0.1737 | 1.84 | 17500 | 0.0833 | 0.0966 | | 0.1713 | 1.9 | 18000 | 0.0799 | 0.0963 | | 0.1703 | 1.95 | 18500 | 0.0802 | 0.0950 | | 0.1735 | 2.0 | 19000 | 0.0785 | 0.0926 | | 0.1619 | 2.06 | 19500 | 0.0785 | 0.0930 | | 0.1707 | 2.11 | 20000 | 0.0787 | 0.0928 | | 0.17 | 2.16 | 20500 | 0.0765 | 0.0902 | | 0.1604 | 2.21 | 21000 | 0.0772 | 0.0918 | | 0.1576 | 2.27 | 21500 | 0.0745 | 0.0912 | | 0.1529 | 2.32 | 22000 | 0.0741 | 0.0906 | | 0.1435 | 2.37 | 22500 | 0.0751 | 0.0888 | | 0.1526 | 2.42 | 23000 | 0.0734 | 0.0892 | | 0.1471 | 2.48 | 23500 | 0.0746 | 0.0886 | | 0.1553 | 2.53 | 24000 | 0.0727 | 0.0872 | | 0.1641 | 2.58 | 24500 | 0.0720 | 0.0862 | | 0.1495 | 2.64 | 25000 | 0.0707 | 0.0868 | | 0.1498 | 2.69 | 25500 | 0.0719 | 0.0864 | | 0.1438 | 2.74 | 26000 | 0.0703 | 0.0853 | | 0.1532 | 2.79 | 26500 | 0.0710 | 0.0854 | | 0.1435 | 2.85 | 27000 | 0.0690 | 0.0847 | | 0.1486 | 2.9 | 27500 | 0.0683 | 0.0882 | | 0.1359 | 2.95 | 28000 | 0.0673 | 0.0839 | | 0.1309 | 3.0 | 28500 | 0.0687 | 0.0843 | | 0.1312 | 3.06 | 29000 | 0.0696 | 0.0865 | | 0.1387 | 3.11 | 29500 | 0.0667 | 0.0857 | | 0.1327 | 3.16 | 30000 | 0.0667 | 0.0845 | | 0.1251 | 3.21 | 30500 | 0.0662 | 0.0820 | | 0.1415 | 3.27 | 31000 | 0.0652 | 0.0831 | | 0.1221 | 3.32 | 31500 | 0.0660 | 0.0822 | | 0.1337 | 3.37 | 32000 | 0.0658 | 0.0799 | | 0.1342 | 3.43 | 32500 | 0.0650 | 0.0808 | | 0.1391 | 3.48 | 33000 | 0.0658 | 0.0791 | | 0.1351 | 3.53 | 33500 | 0.0654 | 0.0794 | | 0.1309 | 3.58 | 34000 | 0.0650 | 0.0781 | | 0.1317 | 3.64 | 34500 | 0.0629 | 0.0783 | | 0.1326 | 3.69 | 35000 | 0.0637 | 0.0795 | | 0.1296 | 3.74 | 35500 | 0.0624 | 0.0773 | | 0.1156 | 3.79 | 36000 | 0.0613 | 0.0759 | | 0.1242 | 3.85 | 36500 | 0.0627 | 0.0761 | | 0.1251 | 3.9 | 37000 | 0.0638 | 0.0758 | | 0.1335 | 3.95 | 37500 | 0.0620 | 0.0756 | | 0.1374 | 4.01 | 38000 | 0.0628 | 0.0756 | | 0.1227 | 4.06 | 38500 | 0.0637 | 0.0770 | | 0.1144 | 4.11 | 39000 | 0.0637 | 0.0775 | | 0.1222 | 4.16 | 39500 | 0.0630 | 0.0738 | | 0.1207 | 4.22 | 40000 | 0.0607 | 0.0720 | | 0.1181 | 4.27 | 40500 | 0.0608 | 0.0724 | | 0.1259 | 4.32 | 41000 | 0.0608 | 0.0734 | | 0.1137 | 4.37 | 41500 | 0.0623 | 0.0718 | | 0.1275 | 4.43 | 42000 | 0.0620 | 0.0721 | | 0.1218 | 4.48 | 42500 | 0.0599 | 0.0703 | | 0.1212 | 4.53 | 43000 | 0.0612 | 0.0708 | | 0.1144 | 4.59 | 43500 | 0.0589 | 0.0702 | | 0.1199 | 4.64 | 44000 | 0.0589 | 0.0695 | | 0.1113 | 4.69 | 44500 | 0.0601 | 0.0698 | | 0.1108 | 4.74 | 45000 | 0.0584 | 0.0695 | | 0.1196 | 4.8 | 45500 | 0.0596 | 0.0694 | | 0.1216 | 4.85 | 46000 | 0.0578 | 0.0703 | | 0.1188 | 4.9 | 46500 | 0.0596 | 0.0684 | | 0.1122 | 4.95 | 47000 | 0.0584 | 0.0671 | | 0.1115 | 5.01 | 47500 | 0.0594 | 0.0682 | | 0.1777 | 5.06 | 48000 | 0.0597 | 0.0682 | | 0.108 | 5.11 | 48500 | 0.0573 | 0.0691 | | 0.1132 | 5.16 | 49000 | 0.0583 | 0.0666 | | 0.1091 | 5.22 | 49500 | 0.0582 | 0.0672 | | 0.1056 | 5.27 | 50000 | 0.0578 | 0.0674 | | 0.1027 | 5.32 | 50500 | 0.0574 | 0.0671 | | 0.1112 | 5.38 | 51000 | 0.0569 | 0.0659 | | 0.1096 | 5.43 | 51500 | 0.0582 | 0.0662 | | 0.1098 | 5.48 | 52000 | 0.0576 | 0.0667 | | 0.1088 | 5.53 | 52500 | 0.0560 | 0.0679 | | 0.1076 | 5.59 | 53000 | 0.0579 | 0.0664 | | 0.1037 | 5.64 | 53500 | 0.0556 | 0.0661 | | 0.1039 | 5.69 | 54000 | 0.0572 | 0.0675 | | 0.108 | 5.74 | 54500 | 0.0562 | 0.0662 | | 0.1069 | 5.8 | 55000 | 0.0576 | 0.0663 | | 0.1066 | 5.85 | 55500 | 0.0564 | 0.0651 | | 0.0939 | 5.9 | 56000 | 0.0566 | 0.0644 | | 0.1118 | 5.96 | 56500 | 0.0570 | 0.0650 | | 0.1111 | 6.01 | 57000 | 0.0563 | 0.0668 | | 0.1014 | 6.06 | 57500 | 0.0557 | 0.0660 | | 0.0971 | 6.11 | 58000 | 0.0567 | 0.0667 | | 0.0932 | 6.17 | 58500 | 0.0559 | 0.0664 | | 0.1002 | 6.22 | 59000 | 0.0551 | 0.0640 | | 0.1028 | 6.27 | 59500 | 0.0560 | 0.0629 | | 0.0992 | 6.32 | 60000 | 0.0547 | 0.0641 | | 0.0975 | 6.38 | 60500 | 0.0556 | 0.0630 | | 0.0957 | 6.43 | 61000 | 0.0555 | 0.0632 | | 0.0931 | 6.48 | 61500 | 0.0546 | 0.0641 | | 0.0999 | 6.54 | 62000 | 0.0556 | 0.0633 | | 0.0998 | 6.59 | 62500 | 0.0539 | 0.0628 | | 0.0991 | 6.64 | 63000 | 0.0559 | 0.0630 | | 0.1027 | 6.69 | 63500 | 0.0549 | 0.0628 | | 0.097 | 6.75 | 64000 | 0.0547 | 0.0628 | | 0.0933 | 6.8 | 64500 | 0.0544 | 0.0633 | | 0.0919 | 6.85 | 65000 | 0.0535 | 0.0640 | | 0.0973 | 6.9 | 65500 | 0.0543 | 0.0619 | | 0.0979 | 6.96 | 66000 | 0.0525 | 0.0620 | | 0.1076 | 7.01 | 66500 | 0.0529 | 0.0615 | | 0.0888 | 7.06 | 67000 | 0.0546 | 0.0617 | | 0.0926 | 7.11 | 67500 | 0.0530 | 0.0636 | | 0.0902 | 7.17 | 68000 | 0.0540 | 0.0631 | | 0.1004 | 7.22 | 68500 | 0.0529 | 0.0624 | | 0.0963 | 7.27 | 69000 | 0.0534 | 0.0631 | | 0.0946 | 7.33 | 69500 | 0.0534 | 0.0601 | | 0.0897 | 7.38 | 70000 | 0.0525 | 0.0607 | | 0.0925 | 7.43 | 70500 | 0.0535 | 0.0599 | | 0.0883 | 7.48 | 71000 | 0.0518 | 0.0605 | | 0.0942 | 7.54 | 71500 | 0.0522 | 0.0587 | | 0.0863 | 7.59 | 72000 | 0.0533 | 0.0593 | | 0.0894 | 7.64 | 72500 | 0.0529 | 0.0587 | | 0.0908 | 7.69 | 73000 | 0.0519 | 0.0596 | | 0.0878 | 7.75 | 73500 | 0.0521 | 0.0585 | | 0.0949 | 7.8 | 74000 | 0.0524 | 0.0588 | | 0.0962 | 7.85 | 74500 | 0.0521 | 0.0581 | | 0.0918 | 7.91 | 75000 | 0.0513 | 0.0579 | | 0.0933 | 7.96 | 75500 | 0.0522 | 0.0582 | | 0.0839 | 8.01 | 76000 | 0.0536 | 0.0579 | | 0.0868 | 8.06 | 76500 | 0.0526 | 0.0577 | | 0.086 | 8.12 | 77000 | 0.0525 | 0.0590 | | 0.0801 | 8.17 | 77500 | 0.0533 | 0.0586 | | 0.0845 | 8.22 | 78000 | 0.0516 | 0.0578 | | 0.0895 | 8.27 | 78500 | 0.0530 | 0.0583 | | 0.0841 | 8.33 | 79000 | 0.0515 | 0.0584 | | 0.0921 | 8.38 | 79500 | 0.0518 | 0.0573 | | 0.0897 | 8.43 | 80000 | 0.0514 | 0.0583 | | 0.0889 | 8.49 | 80500 | 0.0508 | 0.0582 | | 0.1783 | 8.54 | 81000 | 0.0507 | 0.0574 | | 0.0854 | 8.59 | 81500 | 0.0505 | 0.0580 | | 0.0855 | 8.64 | 82000 | 0.0513 | 0.0577 | | 0.0843 | 8.7 | 82500 | 0.0508 | 0.0580 | | 0.0858 | 8.75 | 83000 | 0.0501 | 0.0578 | | 0.0814 | 8.8 | 83500 | 0.0509 | 0.0580 | | 0.0823 | 8.85 | 84000 | 0.0509 | 0.0575 | | 0.0857 | 8.91 | 84500 | 0.0499 | 0.0599 | | 0.0787 | 8.96 | 85000 | 0.0505 | 0.0598 | | 0.0805 | 9.01 | 85500 | 0.0510 | 0.0606 | | 0.0798 | 9.07 | 86000 | 0.0515 | 0.0603 | | 0.0812 | 9.12 | 86500 | 0.0507 | 0.0586 | | 0.0781 | 9.17 | 87000 | 0.0511 | 0.0612 | | 0.0814 | 9.22 | 87500 | 0.0508 | 0.0589 | | 0.0821 | 9.28 | 88000 | 0.0507 | 0.0588 | | 0.0808 | 9.33 | 88500 | 0.0498 | 0.0571 | | 0.0793 | 9.38 | 89000 | 0.0502 | 0.0574 | | 0.0791 | 9.43 | 89500 | 0.0498 | 0.0568 | | 0.0779 | 9.49 | 90000 | 0.0507 | 0.0570 | | 0.0777 | 9.54 | 90500 | 0.0508 | 0.0573 | | 0.0816 | 9.59 | 91000 | 0.0493 | 0.0573 | | 0.0835 | 9.64 | 91500 | 0.0496 | 0.0563 | | 0.0827 | 9.7 | 92000 | 0.0493 | 0.0559 | | 0.0904 | 9.75 | 92500 | 0.0492 | 0.0564 | | 0.0753 | 9.8 | 93000 | 0.0503 | 0.0557 | | 0.0748 | 9.86 | 93500 | 0.0493 | 0.0554 | | 0.0759 | 9.91 | 94000 | 0.0499 | 0.0557 | | 0.0825 | 9.96 | 94500 | 0.0498 | 0.0566 | | 0.0787 | 10.01 | 95000 | 0.0499 | 0.0561 | | 0.0804 | 10.07 | 95500 | 0.0499 | 0.0562 | | 0.0784 | 10.12 | 96000 | 0.0500 | 0.0555 | | 0.0747 | 10.17 | 96500 | 0.0497 | 0.0548 | | 0.0748 | 10.22 | 97000 | 0.0492 | 0.0565 | | 0.0732 | 10.28 | 97500 | 0.0493 | 0.0547 | | 0.0766 | 10.33 | 98000 | 0.0490 | 0.0552 | | 0.0762 | 10.38 | 98500 | 0.0504 | 0.0551 | | 0.0744 | 10.44 | 99000 | 0.0496 | 0.0553 | | 0.0702 | 10.49 | 99500 | 0.0496 | 0.0548 | | 0.0802 | 10.54 | 100000 | 0.0499 | 0.0545 | | 0.1605 | 10.59 | 100500 | 0.0477 | 0.0543 | | 0.0768 | 10.65 | 101000 | 0.0487 | 0.0552 | | 0.0833 | 10.7 | 101500 | 0.0495 | 0.0550 | | 0.0782 | 10.75 | 102000 | 0.0479 | 0.0553 | | 0.0813 | 10.8 | 102500 | 0.0490 | 0.0542 | | 0.0712 | 10.86 | 103000 | 0.0485 | 0.0541 | | 0.0703 | 10.91 | 103500 | 0.0486 | 0.0544 | | 0.0765 | 10.96 | 104000 | 0.0480 | 0.0538 | | 0.0796 | 11.02 | 104500 | 0.0486 | 0.0535 | | 0.0778 | 11.07 | 105000 | 0.0492 | 0.0535 | | 0.0735 | 11.12 | 105500 | 0.0494 | 0.0533 | | 0.068 | 11.17 | 106000 | 0.0485 | 0.0528 | | 0.0687 | 11.23 | 106500 | 0.0498 | 0.0534 | | 0.0641 | 11.28 | 107000 | 0.0493 | 0.0534 | | 0.0712 | 11.33 | 107500 | 0.0485 | 0.0526 | | 0.0827 | 11.38 | 108000 | 0.0484 | 0.0530 | | 0.0715 | 11.44 | 108500 | 0.0480 | 0.0533 | | 0.0733 | 11.49 | 109000 | 0.0482 | 0.0532 | | 0.0754 | 11.54 | 109500 | 0.0481 | 0.0537 | | 0.0719 | 11.59 | 110000 | 0.0475 | 0.0533 | | 0.0707 | 11.65 | 110500 | 0.0479 | 0.0536 | | 0.0687 | 11.7 | 111000 | 0.0483 | 0.0535 | | 0.0713 | 11.75 | 111500 | 0.0485 | 0.0535 | | 0.0674 | 11.81 | 112000 | 0.0482 | 0.0537 | | 0.0704 | 11.86 | 112500 | 0.0487 | 0.0537 | | 0.0691 | 11.91 | 113000 | 0.0484 | 0.0541 | | 0.0708 | 11.96 | 113500 | 0.0485 | 0.0548 | | 0.0683 | 12.02 | 114000 | 0.0487 | 0.0541 | | 0.0691 | 12.07 | 114500 | 0.0492 | 0.0540 | | 0.0679 | 12.12 | 115000 | 0.0486 | 0.0540 | | 0.073 | 12.17 | 115500 | 0.0479 | 0.0545 | | 0.0647 | 12.23 | 116000 | 0.0484 | 0.0534 | | 0.0663 | 12.28 | 116500 | 0.0484 | 0.0532 | | 0.0687 | 12.33 | 117000 | 0.0483 | 0.0532 | | 0.0696 | 12.39 | 117500 | 0.0482 | 0.0541 | | 0.068 | 12.44 | 118000 | 0.0487 | 0.0531 | | 0.0681 | 12.49 | 118500 | 0.0483 | 0.0530 | | 0.0774 | 12.54 | 119000 | 0.0481 | 0.0533 | | 0.0656 | 12.6 | 119500 | 0.0484 | 0.0529 | | 0.0628 | 12.65 | 120000 | 0.0479 | 0.0533 | | 0.0657 | 12.7 | 120500 | 0.0490 | 0.0538 | | 0.0668 | 12.75 | 121000 | 0.0485 | 0.0533 | | 0.0656 | 12.81 | 121500 | 0.0484 | 0.0531 | | 0.0745 | 12.86 | 122000 | 0.0474 | 0.0526 | | 0.0654 | 12.91 | 122500 | 0.0485 | 0.0528 | | 0.0764 | 12.97 | 123000 | 0.0482 | 0.0529 | | 0.0673 | 13.02 | 123500 | 0.0491 | 0.0526 | | 0.0649 | 13.07 | 124000 | 0.0489 | 0.0527 | | 0.0655 | 13.12 | 124500 | 0.0485 | 0.0520 | | 0.0688 | 13.18 | 125000 | 0.0476 | 0.0524 | | 0.0683 | 13.23 | 125500 | 0.0475 | 0.0523 | | 0.0632 | 13.28 | 126000 | 0.0480 | 0.0528 | | 0.063 | 13.33 | 126500 | 0.0483 | 0.0528 | | 0.1418 | 13.39 | 127000 | 0.0464 | 0.0531 | | 0.0693 | 13.44 | 127500 | 0.0473 | 0.0525 | | 0.0696 | 13.49 | 128000 | 0.0477 | 0.0519 | | 0.0644 | 13.54 | 128500 | 0.0477 | 0.0520 | | 0.0625 | 13.6 | 129000 | 0.0480 | 0.0518 | | 0.0682 | 13.65 | 129500 | 0.0471 | 0.0517 | | 0.0698 | 13.7 | 130000 | 0.0480 | 0.0521 | | 0.0643 | 13.76 | 130500 | 0.0482 | 0.0522 | | 0.065 | 13.81 | 131000 | 0.0478 | 0.0521 | | 0.0648 | 13.86 | 131500 | 0.0482 | 0.0519 | | 0.0689 | 13.91 | 132000 | 0.0476 | 0.0520 | | 0.0721 | 13.97 | 132500 | 0.0473 | 0.0523 | | 0.0652 | 14.02 | 133000 | 0.0474 | 0.0519 | | 0.0651 | 14.07 | 133500 | 0.0479 | 0.0519 | | 0.0638 | 14.12 | 134000 | 0.0478 | 0.0520 | | 0.0626 | 14.18 | 134500 | 0.0482 | 0.0519 | | 0.0656 | 14.23 | 135000 | 0.0479 | 0.0521 | | 0.0633 | 14.28 | 135500 | 0.0478 | 0.0519 | | 0.0665 | 14.34 | 136000 | 0.0480 | 0.0519 | | 0.0638 | 14.39 | 136500 | 0.0478 | 0.0517 | | 0.0691 | 14.44 | 137000 | 0.0474 | 0.0515 | | 0.0642 | 14.49 | 137500 | 0.0476 | 0.0514 | | 0.0696 | 14.55 | 138000 | 0.0475 | 0.0515 | | 0.0601 | 14.6 | 138500 | 0.0478 | 0.0515 | | 0.0616 | 14.65 | 139000 | 0.0476 | 0.0515 | | 0.0648 | 14.7 | 139500 | 0.0477 | 0.0516 | | 0.0682 | 14.76 | 140000 | 0.0477 | 0.0515 | | 0.0641 | 14.81 | 140500 | 0.0474 | 0.0515 | | 0.0579 | 14.86 | 141000 | 0.0475 | 0.0514 | | 0.0613 | 14.92 | 141500 | 0.0475 | 0.0514 | | 0.0624 | 14.97 | 142000 | 0.0475 | 0.0514 | ### Framework versions - Transformers 4.18.0 - Pytorch 1.11.0+cu113 - Datasets 2.1.0 - Tokenizers 0.12.1