<!-- 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-03
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: 0.1342
- Wer: 0.0814
- Cer: 0.0252
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 |
---|---|---|---|---|---|
28.7197 | 1.0 | 86 | 7.8136 | 1.0 | 0.9778 |
12.9844 | 2.0 | 172 | 3.2566 | 1.0 | 1.0 |
5.5212 | 3.0 | 258 | 2.9239 | 1.0 | 1.0 |
3.0047 | 4.0 | 344 | 2.9117 | 1.0 | 1.0 |
2.92 | 5.0 | 430 | 2.8873 | 1.0 | 1.0 |
2.8994 | 6.0 | 516 | 2.8767 | 1.0 | 1.0 |
2.8914 | 7.0 | 602 | 2.8744 | 1.0 | 1.0 |
2.8914 | 8.0 | 688 | 2.7465 | 1.0 | 1.0 |
2.8674 | 9.0 | 774 | 1.6305 | 1.0 | 0.5600 |
2.278 | 10.0 | 860 | 0.5303 | 0.3196 | 0.0788 |
1.0525 | 11.0 | 946 | 0.3310 | 0.1926 | 0.0524 |
0.6335 | 12.0 | 1032 | 0.2698 | 0.1597 | 0.0454 |
0.5092 | 13.0 | 1118 | 0.2393 | 0.1413 | 0.0410 |
0.4302 | 14.0 | 1204 | 0.2231 | 0.1275 | 0.0393 |
0.4302 | 15.0 | 1290 | 0.2089 | 0.1227 | 0.0376 |
0.3908 | 16.0 | 1376 | 0.1960 | 0.1146 | 0.0355 |
0.3692 | 17.0 | 1462 | 0.1816 | 0.1038 | 0.0325 |
0.3152 | 18.0 | 1548 | 0.1774 | 0.1067 | 0.0334 |
0.2978 | 19.0 | 1634 | 0.1765 | 0.1007 | 0.0320 |
0.2845 | 20.0 | 1720 | 0.1700 | 0.0981 | 0.0308 |
0.2807 | 21.0 | 1806 | 0.1688 | 0.0998 | 0.0313 |
0.2807 | 22.0 | 1892 | 0.1620 | 0.0984 | 0.0308 |
0.2485 | 23.0 | 1978 | 0.1561 | 0.0962 | 0.0304 |
0.2459 | 24.0 | 2064 | 0.1558 | 0.0950 | 0.0303 |
0.2515 | 25.0 | 2150 | 0.1543 | 0.0948 | 0.0297 |
0.2161 | 26.0 | 2236 | 0.1518 | 0.0948 | 0.0302 |
0.2185 | 27.0 | 2322 | 0.1452 | 0.0921 | 0.0284 |
0.2224 | 28.0 | 2408 | 0.1468 | 0.0912 | 0.0284 |
0.2224 | 29.0 | 2494 | 0.1455 | 0.0926 | 0.0283 |
0.225 | 30.0 | 2580 | 0.1455 | 0.0931 | 0.0285 |
0.2025 | 31.0 | 2666 | 0.1472 | 0.0914 | 0.0282 |
0.2011 | 32.0 | 2752 | 0.1423 | 0.0912 | 0.0278 |
0.1886 | 33.0 | 2838 | 0.1433 | 0.0914 | 0.0282 |
0.1912 | 34.0 | 2924 | 0.1446 | 0.0867 | 0.0281 |
0.1878 | 35.0 | 3010 | 0.1403 | 0.0890 | 0.0276 |
0.1878 | 36.0 | 3096 | 0.1412 | 0.0888 | 0.0275 |
0.1875 | 37.0 | 3182 | 0.1420 | 0.0862 | 0.0274 |
0.1757 | 38.0 | 3268 | 0.1447 | 0.0898 | 0.0283 |
0.1809 | 39.0 | 3354 | 0.1403 | 0.0883 | 0.0274 |
0.185 | 40.0 | 3440 | 0.1406 | 0.0878 | 0.0277 |
0.1773 | 41.0 | 3526 | 0.1397 | 0.0840 | 0.0271 |
0.1607 | 42.0 | 3612 | 0.1403 | 0.0910 | 0.0280 |
0.1607 | 43.0 | 3698 | 0.1424 | 0.0910 | 0.0280 |
0.1686 | 44.0 | 3784 | 0.1401 | 0.0886 | 0.0276 |
0.1651 | 45.0 | 3870 | 0.1403 | 0.0869 | 0.0270 |
0.1494 | 46.0 | 3956 | 0.1419 | 0.0886 | 0.0273 |
0.1665 | 47.0 | 4042 | 0.1429 | 0.0869 | 0.0270 |
0.1591 | 48.0 | 4128 | 0.1397 | 0.0878 | 0.0274 |
0.1426 | 49.0 | 4214 | 0.1416 | 0.0862 | 0.0272 |
0.1496 | 50.0 | 4300 | 0.1406 | 0.0847 | 0.0267 |
0.1496 | 51.0 | 4386 | 0.1414 | 0.0845 | 0.0270 |
0.172 | 52.0 | 4472 | 0.1382 | 0.0855 | 0.0272 |
0.152 | 53.0 | 4558 | 0.1395 | 0.0869 | 0.0271 |
0.1548 | 54.0 | 4644 | 0.1394 | 0.0871 | 0.0272 |
0.1453 | 55.0 | 4730 | 0.1373 | 0.0816 | 0.0255 |
0.1445 | 56.0 | 4816 | 0.1395 | 0.0838 | 0.0264 |
0.1334 | 57.0 | 4902 | 0.1373 | 0.0864 | 0.0267 |
0.1334 | 58.0 | 4988 | 0.1360 | 0.0850 | 0.0265 |
0.1381 | 59.0 | 5074 | 0.1404 | 0.0847 | 0.0265 |
0.1326 | 60.0 | 5160 | 0.1386 | 0.0857 | 0.0266 |
0.1355 | 61.0 | 5246 | 0.1364 | 0.0821 | 0.0263 |
0.1356 | 62.0 | 5332 | 0.1394 | 0.0838 | 0.0263 |
0.1264 | 63.0 | 5418 | 0.1413 | 0.0831 | 0.0260 |
0.1268 | 64.0 | 5504 | 0.1369 | 0.0795 | 0.0254 |
0.1268 | 65.0 | 5590 | 0.1377 | 0.0819 | 0.0259 |
0.1295 | 66.0 | 5676 | 0.1368 | 0.0819 | 0.0259 |
0.1272 | 67.0 | 5762 | 0.1390 | 0.0828 | 0.0259 |
0.1578 | 68.0 | 5848 | 0.1377 | 0.0836 | 0.0260 |
0.1286 | 69.0 | 5934 | 0.1356 | 0.0800 | 0.0252 |
0.1248 | 70.0 | 6020 | 0.1357 | 0.0840 | 0.0262 |
0.12 | 71.0 | 6106 | 0.1367 | 0.0819 | 0.0259 |
0.12 | 72.0 | 6192 | 0.1386 | 0.0824 | 0.0256 |
0.1268 | 73.0 | 6278 | 0.1358 | 0.0838 | 0.0261 |
0.1246 | 74.0 | 6364 | 0.1368 | 0.0831 | 0.0260 |
0.117 | 75.0 | 6450 | 0.1356 | 0.0824 | 0.0255 |
0.1256 | 76.0 | 6536 | 0.1355 | 0.0816 | 0.0253 |
0.1361 | 77.0 | 6622 | 0.1358 | 0.0836 | 0.0259 |
0.1158 | 78.0 | 6708 | 0.1356 | 0.0807 | 0.0255 |
0.1158 | 79.0 | 6794 | 0.1360 | 0.0788 | 0.0248 |
0.1338 | 80.0 | 6880 | 0.1360 | 0.0793 | 0.0253 |
0.117 | 81.0 | 6966 | 0.1352 | 0.0785 | 0.0248 |
0.1123 | 82.0 | 7052 | 0.1342 | 0.0814 | 0.0252 |
0.1232 | 83.0 | 7138 | 0.1357 | 0.0831 | 0.0256 |
0.1178 | 84.0 | 7224 | 0.1348 | 0.0800 | 0.0252 |
0.1151 | 85.0 | 7310 | 0.1354 | 0.0819 | 0.0253 |
0.1151 | 86.0 | 7396 | 0.1350 | 0.0804 | 0.0251 |
0.1211 | 87.0 | 7482 | 0.1365 | 0.0816 | 0.0253 |
0.1079 | 88.0 | 7568 | 0.1359 | 0.0802 | 0.0252 |
0.1201 | 89.0 | 7654 | 0.1368 | 0.0833 | 0.0254 |
0.1073 | 90.0 | 7740 | 0.1361 | 0.0819 | 0.0252 |
0.1088 | 91.0 | 7826 | 0.1358 | 0.0819 | 0.0252 |
0.1156 | 92.0 | 7912 | 0.1354 | 0.0804 | 0.0251 |
0.1156 | 93.0 | 7998 | 0.1358 | 0.0802 | 0.0249 |
0.113 | 94.0 | 8084 | 0.1357 | 0.0795 | 0.0249 |
0.1138 | 95.0 | 8170 | 0.1357 | 0.0812 | 0.0251 |
0.1144 | 96.0 | 8256 | 0.1352 | 0.0812 | 0.0251 |
0.1041 | 97.0 | 8342 | 0.1350 | 0.0802 | 0.0249 |
0.1103 | 98.0 | 8428 | 0.1353 | 0.0809 | 0.0250 |
0.1056 | 99.0 | 8514 | 0.1349 | 0.0812 | 0.0251 |
0.1019 | 100.0 | 8600 | 0.1349 | 0.0804 | 0.0250 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3