Model: alinerodrigues/wav2vec2-large-xlsr-mecita-coraa-portuguese-2-all-03 Source: Original Platform
license, tags, metrics, model-index
| license | tags | metrics | model-index | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| apache-2.0 |
|
|
|
wav2vec2-large-xlsr-mecita-coraa-portuguese-2-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: 0.1350
- Wer: 0.0841
- Cer: 0.0261
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 |
|---|---|---|---|---|---|
| 28.65 | 1.0 | 86 | 3.1940 | 1.0 | 1.0 |
| 8.1771 | 2.0 | 172 | 2.9853 | 1.0 | 1.0 |
| 3.0358 | 3.0 | 258 | 2.8955 | 1.0 | 1.0 |
| 2.9359 | 4.0 | 344 | 2.9086 | 1.0 | 1.0 |
| 2.9026 | 5.0 | 430 | 2.6946 | 0.9998 | 0.9998 |
| 2.4934 | 6.0 | 516 | 0.9106 | 0.4944 | 0.1257 |
| 1.1576 | 7.0 | 602 | 0.4618 | 0.2291 | 0.0627 |
| 1.1576 | 8.0 | 688 | 0.3427 | 0.1871 | 0.0538 |
| 0.7498 | 9.0 | 774 | 0.2838 | 0.1673 | 0.0480 |
| 0.5493 | 10.0 | 860 | 0.2504 | 0.1490 | 0.0442 |
| 0.4735 | 11.0 | 946 | 0.2318 | 0.1388 | 0.0412 |
| 0.4183 | 12.0 | 1032 | 0.2174 | 0.1215 | 0.0374 |
| 0.3924 | 13.0 | 1118 | 0.2019 | 0.1127 | 0.0357 |
| 0.3497 | 14.0 | 1204 | 0.1876 | 0.1112 | 0.0343 |
| 0.3497 | 15.0 | 1290 | 0.1865 | 0.1110 | 0.0351 |
| 0.3335 | 16.0 | 1376 | 0.1759 | 0.1065 | 0.0328 |
| 0.3259 | 17.0 | 1462 | 0.1718 | 0.0941 | 0.0310 |
| 0.2822 | 18.0 | 1548 | 0.1664 | 0.0965 | 0.0307 |
| 0.2689 | 19.0 | 1634 | 0.1638 | 0.0951 | 0.0302 |
| 0.262 | 20.0 | 1720 | 0.1616 | 0.0937 | 0.0307 |
| 0.2623 | 21.0 | 1806 | 0.1574 | 0.0927 | 0.0299 |
| 0.2623 | 22.0 | 1892 | 0.1559 | 0.0944 | 0.0306 |
| 0.2344 | 23.0 | 1978 | 0.1538 | 0.0920 | 0.0293 |
| 0.2315 | 24.0 | 2064 | 0.1533 | 0.0927 | 0.0298 |
| 0.2411 | 25.0 | 2150 | 0.1520 | 0.0908 | 0.0294 |
| 0.2102 | 26.0 | 2236 | 0.1496 | 0.0922 | 0.0297 |
| 0.2115 | 27.0 | 2322 | 0.1473 | 0.0979 | 0.0298 |
| 0.2139 | 28.0 | 2408 | 0.1496 | 0.0958 | 0.0300 |
| 0.2139 | 29.0 | 2494 | 0.1446 | 0.0903 | 0.0289 |
| 0.2218 | 30.0 | 2580 | 0.1482 | 0.0913 | 0.0291 |
| 0.1982 | 31.0 | 2666 | 0.1439 | 0.0891 | 0.0284 |
| 0.196 | 32.0 | 2752 | 0.1428 | 0.0934 | 0.0286 |
| 0.1846 | 33.0 | 2838 | 0.1404 | 0.0906 | 0.0281 |
| 0.1873 | 34.0 | 2924 | 0.1395 | 0.0856 | 0.0272 |
| 0.1854 | 35.0 | 3010 | 0.1354 | 0.0846 | 0.0268 |
| 0.1854 | 36.0 | 3096 | 0.1359 | 0.0858 | 0.0271 |
| 0.1851 | 37.0 | 3182 | 0.1356 | 0.0903 | 0.0276 |
| 0.1733 | 38.0 | 3268 | 0.1380 | 0.0851 | 0.0270 |
| 0.1781 | 39.0 | 3354 | 0.1359 | 0.0813 | 0.0258 |
| 0.1821 | 40.0 | 3440 | 0.1383 | 0.0863 | 0.0274 |
| 0.1724 | 41.0 | 3526 | 0.1377 | 0.0822 | 0.0265 |
| 0.1591 | 42.0 | 3612 | 0.1379 | 0.0844 | 0.0266 |
| 0.1591 | 43.0 | 3698 | 0.1350 | 0.0841 | 0.0261 |
| 0.168 | 44.0 | 3784 | 0.1362 | 0.0846 | 0.0265 |
| 0.162 | 45.0 | 3870 | 0.1353 | 0.0820 | 0.0266 |
| 0.145 | 46.0 | 3956 | 0.1414 | 0.0849 | 0.0270 |
| 0.1639 | 47.0 | 4042 | 0.1400 | 0.0837 | 0.0266 |
| 0.1572 | 48.0 | 4128 | 0.1387 | 0.0815 | 0.0261 |
| 0.14 | 49.0 | 4214 | 0.1384 | 0.0822 | 0.0258 |
| 0.1473 | 50.0 | 4300 | 0.1399 | 0.0827 | 0.0265 |
| 0.1473 | 51.0 | 4386 | 0.1408 | 0.0822 | 0.0265 |
| 0.1722 | 52.0 | 4472 | 0.1411 | 0.0851 | 0.0267 |
| 0.1518 | 53.0 | 4558 | 0.1410 | 0.0860 | 0.0273 |
| 0.1506 | 54.0 | 4644 | 0.1403 | 0.0858 | 0.0270 |
| 0.1442 | 55.0 | 4730 | 0.1380 | 0.0832 | 0.0262 |
| 0.1427 | 56.0 | 4816 | 0.1395 | 0.0815 | 0.0265 |
| 0.132 | 57.0 | 4902 | 0.1384 | 0.0813 | 0.0259 |
| 0.132 | 58.0 | 4988 | 0.1353 | 0.0818 | 0.0261 |
| 0.1374 | 59.0 | 5074 | 0.1405 | 0.0834 | 0.0265 |
| 0.1318 | 60.0 | 5160 | 0.1383 | 0.0811 | 0.0258 |
| 0.1339 | 61.0 | 5246 | 0.1386 | 0.0813 | 0.0261 |
| 0.1346 | 62.0 | 5332 | 0.1399 | 0.0801 | 0.0253 |
| 0.1261 | 63.0 | 5418 | 0.1430 | 0.0832 | 0.0263 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.1.1+cu121
- Datasets 2.15.0
- Tokenizers 0.13.3
Description
Model synced from source: alinerodrigues/wav2vec2-large-xlsr-mecita-coraa-portuguese-2-all-03