<!-- 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-09
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: 0.2055
- Wer: 0.1135
- Cer: 0.0374
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 |
---|---|---|---|---|---|
27.8255 | 1.0 | 86 | 6.4686 | 1.0 | 1.0 |
9.4861 | 2.0 | 172 | 6.1452 | 0.9703 | 0.9624 |
4.0433 | 3.0 | 258 | 6.5288 | 0.9552 | 0.9417 |
3.757 | 4.0 | 344 | 6.4359 | 1.0 | 1.0 |
3.5579 | 5.0 | 430 | 6.5739 | 1.0 | 1.0 |
3.4195 | 6.0 | 516 | 6.5713 | 0.9768 | 0.9849 |
3.6813 | 7.0 | 602 | 6.5877 | 0.9749 | 0.9137 |
3.6813 | 8.0 | 688 | 6.6049 | 0.9588 | 0.8732 |
3.3744 | 9.0 | 774 | 6.5856 | 0.9770 | 0.8666 |
3.5373 | 10.0 | 860 | 6.6468 | 0.9583 | 0.8608 |
3.1637 | 11.0 | 946 | 6.6958 | 0.9720 | 0.8476 |
3.8044 | 12.0 | 1032 | 6.6167 | 0.9703 | 0.8258 |
2.8859 | 13.0 | 1118 | 6.4369 | 0.9773 | 0.7946 |
3.4141 | 14.0 | 1204 | 6.4624 | 0.9770 | 0.7793 |
3.4141 | 15.0 | 1290 | 6.4052 | 0.9775 | 0.7566 |
3.4378 | 16.0 | 1376 | 6.3971 | 0.9787 | 0.7434 |
3.1892 | 17.0 | 1462 | 6.2175 | 0.9739 | 0.7629 |
3.1787 | 18.0 | 1548 | 4.7706 | 0.9739 | 0.7491 |
3.3283 | 19.0 | 1634 | 6.2475 | 0.9823 | 0.7578 |
3.0856 | 20.0 | 1720 | 5.3353 | 0.9698 | 0.7202 |
3.1698 | 21.0 | 1806 | 5.3445 | 0.9703 | 0.7256 |
3.1698 | 22.0 | 1892 | 3.5694 | 0.9749 | 0.7154 |
3.0717 | 23.0 | 1978 | 4.7239 | 0.9739 | 0.7159 |
3.142 | 24.0 | 2064 | 3.2030 | 0.9722 | 0.7019 |
2.9683 | 25.0 | 2150 | 4.7049 | 0.9622 | 0.7023 |
3.135 | 26.0 | 2236 | 3.2585 | 0.9605 | 0.6932 |
3.1443 | 27.0 | 2322 | 2.8654 | 0.9655 | 0.6901 |
2.7771 | 28.0 | 2408 | 3.3484 | 0.9693 | 0.6844 |
2.7771 | 29.0 | 2494 | 2.5680 | 0.9607 | 0.6808 |
2.7258 | 30.0 | 2580 | 2.6370 | 0.9634 | 0.6768 |
2.8003 | 31.0 | 2666 | 2.4710 | 0.9636 | 0.6717 |
2.8051 | 32.0 | 2752 | 2.6271 | 0.9605 | 0.6689 |
2.5177 | 33.0 | 2838 | 2.3597 | 0.9509 | 0.6670 |
2.7207 | 34.0 | 2924 | 2.3109 | 0.9571 | 0.6564 |
2.4472 | 35.0 | 3010 | 2.3818 | 0.9646 | 0.6348 |
2.4472 | 36.0 | 3096 | 2.2100 | 0.9514 | 0.6569 |
2.6165 | 37.0 | 3182 | 2.1373 | 0.9526 | 0.6497 |
2.4184 | 38.0 | 3268 | 2.0397 | 0.9459 | 0.6180 |
2.5079 | 39.0 | 3354 | 1.8301 | 0.9349 | 0.5742 |
2.3417 | 40.0 | 3440 | 1.5497 | 0.9004 | 0.4828 |
1.853 | 41.0 | 3526 | 1.2096 | 0.7804 | 0.3135 |
1.4946 | 42.0 | 3612 | 0.8933 | 0.7035 | 0.2430 |
1.4946 | 43.0 | 3698 | 0.7703 | 0.5532 | 0.1646 |
1.1791 | 44.0 | 3784 | 0.5455 | 0.4614 | 0.1339 |
0.9583 | 45.0 | 3870 | 0.4740 | 0.3726 | 0.1059 |
0.8472 | 46.0 | 3956 | 0.4110 | 0.3099 | 0.0878 |
0.6646 | 47.0 | 4042 | 0.3723 | 0.2900 | 0.0796 |
0.6322 | 48.0 | 4128 | 0.3474 | 0.2493 | 0.0701 |
0.5803 | 49.0 | 4214 | 0.3237 | 0.2404 | 0.0681 |
0.5056 | 50.0 | 4300 | 0.3052 | 0.2079 | 0.0619 |
0.5056 | 51.0 | 4386 | 0.2910 | 0.1892 | 0.0573 |
0.4646 | 52.0 | 4472 | 0.2821 | 0.1796 | 0.0542 |
0.4439 | 53.0 | 4558 | 0.2698 | 0.1578 | 0.0497 |
0.3897 | 54.0 | 4644 | 0.2598 | 0.1573 | 0.0488 |
0.4143 | 55.0 | 4730 | 0.2558 | 0.1456 | 0.0468 |
0.3624 | 56.0 | 4816 | 0.2467 | 0.1487 | 0.0465 |
0.3814 | 57.0 | 4902 | 0.2456 | 0.1408 | 0.0449 |
0.3814 | 58.0 | 4988 | 0.2434 | 0.1401 | 0.0445 |
0.3546 | 59.0 | 5074 | 0.2400 | 0.1315 | 0.0426 |
0.3368 | 60.0 | 5160 | 0.2426 | 0.1329 | 0.0427 |
0.3561 | 61.0 | 5246 | 0.2380 | 0.1298 | 0.0420 |
0.3077 | 62.0 | 5332 | 0.2306 | 0.1284 | 0.0417 |
0.2969 | 63.0 | 5418 | 0.2290 | 0.1257 | 0.0411 |
0.2857 | 64.0 | 5504 | 0.2220 | 0.1226 | 0.0398 |
0.2857 | 65.0 | 5590 | 0.2245 | 0.1262 | 0.0411 |
0.2834 | 66.0 | 5676 | 0.2223 | 0.1238 | 0.0399 |
0.3022 | 67.0 | 5762 | 0.2174 | 0.1226 | 0.0397 |
0.2479 | 68.0 | 5848 | 0.2239 | 0.1226 | 0.0395 |
0.2648 | 69.0 | 5934 | 0.2193 | 0.1195 | 0.0384 |
0.2546 | 70.0 | 6020 | 0.2124 | 0.1212 | 0.0388 |
0.2645 | 71.0 | 6106 | 0.2175 | 0.1219 | 0.0394 |
0.2645 | 72.0 | 6192 | 0.2135 | 0.1195 | 0.0383 |
0.2397 | 73.0 | 6278 | 0.2119 | 0.1202 | 0.0385 |
0.2508 | 74.0 | 6364 | 0.2118 | 0.1157 | 0.0385 |
0.2588 | 75.0 | 6450 | 0.2109 | 0.1207 | 0.0385 |
0.2556 | 76.0 | 6536 | 0.2090 | 0.1123 | 0.0376 |
0.2376 | 77.0 | 6622 | 0.2080 | 0.1135 | 0.0378 |
0.2441 | 78.0 | 6708 | 0.2100 | 0.1133 | 0.0377 |
0.2441 | 79.0 | 6794 | 0.2090 | 0.1133 | 0.0380 |
0.2158 | 80.0 | 6880 | 0.2075 | 0.1140 | 0.0377 |
0.227 | 81.0 | 6966 | 0.2081 | 0.1142 | 0.0375 |
0.2196 | 82.0 | 7052 | 0.2088 | 0.1135 | 0.0377 |
0.2266 | 83.0 | 7138 | 0.2127 | 0.1116 | 0.0371 |
0.2055 | 84.0 | 7224 | 0.2087 | 0.1128 | 0.0373 |
0.22 | 85.0 | 7310 | 0.2062 | 0.1147 | 0.0376 |
0.22 | 86.0 | 7396 | 0.2059 | 0.1137 | 0.0374 |
0.2055 | 87.0 | 7482 | 0.2074 | 0.1149 | 0.0376 |
0.2282 | 88.0 | 7568 | 0.2055 | 0.1135 | 0.0374 |
0.2266 | 89.0 | 7654 | 0.2078 | 0.1090 | 0.0367 |
0.2054 | 90.0 | 7740 | 0.2062 | 0.1111 | 0.0371 |
0.2164 | 91.0 | 7826 | 0.2058 | 0.1125 | 0.0373 |
0.2047 | 92.0 | 7912 | 0.2087 | 0.1078 | 0.0368 |
0.2047 | 93.0 | 7998 | 0.2074 | 0.1075 | 0.0363 |
0.2125 | 94.0 | 8084 | 0.2068 | 0.1099 | 0.0370 |
0.213 | 95.0 | 8170 | 0.2064 | 0.1102 | 0.0367 |
0.1905 | 96.0 | 8256 | 0.2056 | 0.1099 | 0.0366 |
0.1947 | 97.0 | 8342 | 0.2063 | 0.1085 | 0.0364 |
0.2015 | 98.0 | 8428 | 0.2064 | 0.1094 | 0.0366 |
0.2028 | 99.0 | 8514 | 0.2062 | 0.1099 | 0.0368 |
0.2001 | 100.0 | 8600 | 0.2063 | 0.1097 | 0.0368 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3