<!-- 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-05
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: 0.2174
- Wer: 0.1429
- Cer: 0.0430
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 |
---|---|---|---|---|---|
28.1646 | 1.0 | 86 | 4.8610 | 0.9832 | 0.9803 |
9.4454 | 2.0 | 172 | 3.3327 | 0.9567 | 0.9034 |
4.3702 | 3.0 | 258 | 3.1962 | 1.0 | 1.0 |
4.398 | 4.0 | 344 | 2.9906 | 1.0 | 1.0 |
3.7462 | 5.0 | 430 | 2.9604 | 1.0 | 1.0 |
3.9079 | 6.0 | 516 | 2.9447 | 0.9659 | 0.9651 |
3.9602 | 7.0 | 602 | 2.8960 | 0.9591 | 0.9286 |
3.9602 | 8.0 | 688 | 2.8418 | 0.9555 | 0.8580 |
3.7183 | 9.0 | 774 | 2.8463 | 0.9516 | 0.8567 |
3.7229 | 10.0 | 860 | 2.9000 | 0.9706 | 0.8790 |
3.7056 | 11.0 | 946 | 2.7891 | 0.9747 | 0.8368 |
4.3909 | 12.0 | 1032 | 2.8128 | 0.9681 | 0.8091 |
3.2494 | 13.0 | 1118 | 2.7410 | 0.9776 | 0.7792 |
4.1417 | 14.0 | 1204 | 2.7921 | 0.9781 | 0.8052 |
4.1417 | 15.0 | 1290 | 2.7801 | 0.9720 | 0.7893 |
3.5813 | 16.0 | 1376 | 2.7245 | 0.9735 | 0.7797 |
3.7739 | 17.0 | 1462 | 2.7206 | 0.9633 | 0.7592 |
3.2666 | 18.0 | 1548 | 2.6764 | 0.9786 | 0.7518 |
3.5191 | 19.0 | 1634 | 2.6704 | 0.9771 | 0.7560 |
4.0333 | 20.0 | 1720 | 2.6269 | 0.9757 | 0.7216 |
3.155 | 21.0 | 1806 | 2.6349 | 0.9710 | 0.7388 |
3.155 | 22.0 | 1892 | 2.7308 | 0.9557 | 0.7501 |
3.4096 | 23.0 | 1978 | 2.5728 | 0.9727 | 0.7167 |
3.6189 | 24.0 | 2064 | 2.6209 | 0.9706 | 0.7134 |
3.1681 | 25.0 | 2150 | 2.5653 | 0.9662 | 0.7091 |
2.851 | 26.0 | 2236 | 2.5492 | 0.9601 | 0.7133 |
3.0625 | 27.0 | 2322 | 2.4696 | 0.9584 | 0.7059 |
2.6773 | 28.0 | 2408 | 2.4480 | 0.9708 | 0.6835 |
2.6773 | 29.0 | 2494 | 2.4551 | 0.9676 | 0.6831 |
2.6704 | 30.0 | 2580 | 2.3923 | 0.9615 | 0.6865 |
2.9334 | 31.0 | 2666 | 2.3834 | 0.9620 | 0.6748 |
2.8201 | 32.0 | 2752 | 2.3488 | 0.9615 | 0.6599 |
2.4969 | 33.0 | 2838 | 2.2633 | 0.9637 | 0.6578 |
2.5507 | 34.0 | 2924 | 2.2478 | 0.9744 | 0.6386 |
2.4824 | 35.0 | 3010 | 2.1524 | 0.9589 | 0.6350 |
2.4824 | 36.0 | 3096 | 2.1081 | 0.9523 | 0.6110 |
2.3294 | 37.0 | 3182 | 2.0454 | 0.9486 | 0.6230 |
2.2776 | 38.0 | 3268 | 1.9534 | 0.9435 | 0.5944 |
2.2266 | 39.0 | 3354 | 1.8659 | 0.9389 | 0.5765 |
2.2176 | 40.0 | 3440 | 1.8386 | 0.9277 | 0.5618 |
2.1509 | 41.0 | 3526 | 1.7009 | 0.9031 | 0.5146 |
2.0032 | 42.0 | 3612 | 1.5778 | 0.8936 | 0.5060 |
2.0032 | 43.0 | 3698 | 1.4811 | 0.8825 | 0.4663 |
1.8864 | 44.0 | 3784 | 1.2900 | 0.8576 | 0.4236 |
1.8023 | 45.0 | 3870 | 1.1809 | 0.7851 | 0.3400 |
1.7414 | 46.0 | 3956 | 1.0532 | 0.7802 | 0.3238 |
1.5595 | 47.0 | 4042 | 0.9481 | 0.7206 | 0.2759 |
1.4958 | 48.0 | 4128 | 0.8725 | 0.6960 | 0.2557 |
1.3644 | 49.0 | 4214 | 0.8044 | 0.6486 | 0.2255 |
1.4496 | 50.0 | 4300 | 0.6878 | 0.6074 | 0.1988 |
1.4496 | 51.0 | 4386 | 0.6003 | 0.5330 | 0.1622 |
1.0768 | 52.0 | 4472 | 0.5278 | 0.4609 | 0.1385 |
0.9745 | 53.0 | 4558 | 0.4708 | 0.3923 | 0.1158 |
0.8776 | 54.0 | 4644 | 0.4379 | 0.3558 | 0.1038 |
0.8135 | 55.0 | 4730 | 0.4097 | 0.3234 | 0.0945 |
0.7573 | 56.0 | 4816 | 0.3926 | 0.3088 | 0.0874 |
0.6982 | 57.0 | 4902 | 0.3704 | 0.2801 | 0.0801 |
0.6982 | 58.0 | 4988 | 0.3488 | 0.2587 | 0.0748 |
0.6478 | 59.0 | 5074 | 0.3264 | 0.2541 | 0.0710 |
0.6391 | 60.0 | 5160 | 0.3236 | 0.2314 | 0.0664 |
0.5856 | 61.0 | 5246 | 0.3106 | 0.2283 | 0.0644 |
0.5626 | 62.0 | 5332 | 0.3044 | 0.2176 | 0.0621 |
0.5698 | 63.0 | 5418 | 0.2993 | 0.2127 | 0.0601 |
0.5238 | 64.0 | 5504 | 0.2926 | 0.2039 | 0.0585 |
0.5238 | 65.0 | 5590 | 0.2852 | 0.2047 | 0.0577 |
0.4932 | 66.0 | 5676 | 0.2698 | 0.1932 | 0.0558 |
0.446 | 67.0 | 5762 | 0.2602 | 0.1932 | 0.0541 |
0.464 | 68.0 | 5848 | 0.2590 | 0.1852 | 0.0531 |
0.4482 | 69.0 | 5934 | 0.2508 | 0.1813 | 0.0520 |
0.4378 | 70.0 | 6020 | 0.2578 | 0.1691 | 0.0500 |
0.4682 | 71.0 | 6106 | 0.2496 | 0.1757 | 0.0505 |
0.4682 | 72.0 | 6192 | 0.2445 | 0.1713 | 0.0492 |
0.4296 | 73.0 | 6278 | 0.2445 | 0.1640 | 0.0483 |
0.4032 | 74.0 | 6364 | 0.2406 | 0.1640 | 0.0485 |
0.4357 | 75.0 | 6450 | 0.2398 | 0.1643 | 0.0484 |
0.4126 | 76.0 | 6536 | 0.2398 | 0.1611 | 0.0480 |
0.4006 | 77.0 | 6622 | 0.2343 | 0.1572 | 0.0468 |
0.3973 | 78.0 | 6708 | 0.2310 | 0.1531 | 0.0464 |
0.3973 | 79.0 | 6794 | 0.2368 | 0.1521 | 0.0459 |
0.3795 | 80.0 | 6880 | 0.2318 | 0.1523 | 0.0457 |
0.3926 | 81.0 | 6966 | 0.2287 | 0.1509 | 0.0455 |
0.3836 | 82.0 | 7052 | 0.2277 | 0.1519 | 0.0456 |
0.3746 | 83.0 | 7138 | 0.2245 | 0.1497 | 0.0445 |
0.3499 | 84.0 | 7224 | 0.2234 | 0.1514 | 0.0445 |
0.3572 | 85.0 | 7310 | 0.2234 | 0.1460 | 0.0444 |
0.3572 | 86.0 | 7396 | 0.2207 | 0.1470 | 0.0442 |
0.3358 | 87.0 | 7482 | 0.2186 | 0.1431 | 0.0441 |
0.3686 | 88.0 | 7568 | 0.2199 | 0.1487 | 0.0444 |
0.3572 | 89.0 | 7654 | 0.2194 | 0.1441 | 0.0442 |
0.3535 | 90.0 | 7740 | 0.2191 | 0.1453 | 0.0443 |
0.334 | 91.0 | 7826 | 0.2181 | 0.1443 | 0.0441 |
0.3153 | 92.0 | 7912 | 0.2177 | 0.1436 | 0.0438 |
0.3153 | 93.0 | 7998 | 0.2174 | 0.1412 | 0.0434 |
0.33 | 94.0 | 8084 | 0.2182 | 0.1455 | 0.0436 |
0.3167 | 95.0 | 8170 | 0.2181 | 0.1429 | 0.0432 |
0.3217 | 96.0 | 8256 | 0.2174 | 0.1429 | 0.0430 |
0.31 | 97.0 | 8342 | 0.2180 | 0.1394 | 0.0428 |
0.3307 | 98.0 | 8428 | 0.2179 | 0.1424 | 0.0431 |
0.3392 | 99.0 | 8514 | 0.2182 | 0.1412 | 0.0429 |
0.3468 | 100.0 | 8600 | 0.2177 | 0.1399 | 0.0430 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3