<!-- 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-grade-4
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: 1.5957
- Wer: 0.9934
- Cer: 0.4456
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 |
|---|---|---|---|---|---|
| 13.7726 | 0.93 | 7 | 11.6190 | 0.9956 | 0.8278 |
| 13.7726 | 2.0 | 15 | 8.5051 | 1.0 | 1.0 |
| 13.7726 | 2.93 | 22 | 6.2165 | 1.0 | 1.0 |
| 13.7726 | 4.0 | 30 | 4.2514 | 1.0 | 1.0 |
| 13.7726 | 4.93 | 37 | 3.6613 | 1.0 | 1.0 |
| 13.7726 | 6.0 | 45 | 3.4125 | 1.0 | 1.0 |
| 13.7726 | 6.93 | 52 | 3.3004 | 1.0 | 1.0 |
| 13.7726 | 8.0 | 60 | 3.1753 | 1.0 | 1.0 |
| 13.7726 | 8.93 | 67 | 3.1030 | 1.0 | 1.0 |
| 13.7726 | 10.0 | 75 | 3.0753 | 1.0 | 1.0 |
| 13.7726 | 10.93 | 82 | 3.0253 | 1.0 | 1.0 |
| 13.7726 | 12.0 | 90 | 3.0037 | 1.0 | 1.0 |
| 13.7726 | 12.93 | 97 | 2.9845 | 1.0 | 1.0 |
| 5.0373 | 14.0 | 105 | 2.9577 | 1.0 | 1.0 |
| 5.0373 | 14.93 | 112 | 2.9440 | 1.0 | 1.0 |
| 5.0373 | 16.0 | 120 | 2.9336 | 1.0 | 1.0 |
| 5.0373 | 16.93 | 127 | 2.9166 | 1.0 | 1.0 |
| 5.0373 | 18.0 | 135 | 2.9040 | 1.0 | 1.0 |
| 5.0373 | 18.93 | 142 | 2.8956 | 1.0 | 1.0 |
| 5.0373 | 20.0 | 150 | 2.8844 | 1.0 | 1.0 |
| 5.0373 | 20.93 | 157 | 2.8762 | 1.0 | 1.0 |
| 5.0373 | 22.0 | 165 | 2.8821 | 1.0 | 1.0 |
| 5.0373 | 22.93 | 172 | 2.8660 | 1.0 | 1.0 |
| 5.0373 | 24.0 | 180 | 2.8739 | 1.0 | 1.0 |
| 5.0373 | 24.93 | 187 | 2.8632 | 1.0 | 1.0 |
| 5.0373 | 26.0 | 195 | 2.8580 | 1.0 | 1.0 |
| 2.8886 | 26.93 | 202 | 2.8565 | 1.0 | 1.0 |
| 2.8886 | 28.0 | 210 | 2.8717 | 1.0 | 1.0 |
| 2.8886 | 28.93 | 217 | 2.8518 | 1.0 | 1.0 |
| 2.8886 | 30.0 | 225 | 2.8849 | 1.0 | 1.0 |
| 2.8886 | 30.93 | 232 | 2.8578 | 1.0 | 1.0 |
| 2.8886 | 32.0 | 240 | 2.8782 | 1.0 | 1.0 |
| 2.8886 | 32.93 | 247 | 2.8465 | 1.0 | 1.0 |
| 2.8886 | 34.0 | 255 | 2.8644 | 1.0 | 1.0 |
| 2.8886 | 34.93 | 262 | 2.8438 | 1.0 | 1.0 |
| 2.8886 | 36.0 | 270 | 2.8466 | 1.0 | 1.0 |
| 2.8886 | 36.93 | 277 | 2.8473 | 1.0 | 1.0 |
| 2.8886 | 38.0 | 285 | 2.8414 | 1.0 | 1.0 |
| 2.8886 | 38.93 | 292 | 2.8444 | 1.0 | 1.0 |
| 2.831 | 40.0 | 300 | 2.8455 | 1.0 | 1.0 |
| 2.831 | 40.93 | 307 | 2.8357 | 1.0 | 1.0 |
| 2.831 | 42.0 | 315 | 2.8320 | 1.0 | 1.0 |
| 2.831 | 42.93 | 322 | 2.8415 | 1.0 | 1.0 |
| 2.831 | 44.0 | 330 | 2.8347 | 1.0 | 1.0 |
| 2.831 | 44.93 | 337 | 2.8386 | 1.0 | 1.0 |
| 2.831 | 46.0 | 345 | 2.8278 | 1.0 | 1.0 |
| 2.831 | 46.93 | 352 | 2.8324 | 1.0 | 1.0 |
| 2.831 | 48.0 | 360 | 2.8290 | 1.0 | 1.0 |
| 2.831 | 48.93 | 367 | 2.8319 | 1.0 | 1.0 |
| 2.831 | 50.0 | 375 | 2.8225 | 1.0 | 1.0 |
| 2.831 | 50.93 | 382 | 2.8048 | 1.0 | 1.0 |
| 2.831 | 52.0 | 390 | 2.8062 | 1.0 | 1.0 |
| 2.831 | 52.93 | 397 | 2.7941 | 1.0 | 1.0 |
| 2.8044 | 54.0 | 405 | 2.7786 | 1.0 | 0.9996 |
| 2.8044 | 54.93 | 412 | 2.7615 | 1.0 | 0.9993 |
| 2.8044 | 56.0 | 420 | 2.7564 | 1.0 | 0.9996 |
| 2.8044 | 56.93 | 427 | 2.7243 | 1.0 | 0.9982 |
| 2.8044 | 58.0 | 435 | 2.7148 | 1.0 | 0.9975 |
| 2.8044 | 58.93 | 442 | 2.6886 | 1.0 | 0.9942 |
| 2.8044 | 60.0 | 450 | 2.6476 | 1.0 | 0.9920 |
| 2.8044 | 60.93 | 457 | 2.6304 | 1.0 | 0.9946 |
| 2.8044 | 62.0 | 465 | 2.5871 | 1.0 | 0.9931 |
| 2.8044 | 62.93 | 472 | 2.5719 | 1.0 | 0.9939 |
| 2.8044 | 64.0 | 480 | 2.5189 | 1.0 | 0.9920 |
| 2.8044 | 64.93 | 487 | 2.4960 | 0.9978 | 0.9910 |
| 2.8044 | 66.0 | 495 | 2.4538 | 1.0 | 0.9808 |
| 2.6802 | 66.93 | 502 | 2.4214 | 1.0 | 0.9693 |
| 2.6802 | 68.0 | 510 | 2.3789 | 1.0 | 0.9407 |
| 2.6802 | 68.93 | 517 | 2.3391 | 0.9978 | 0.9212 |
| 2.6802 | 70.0 | 525 | 2.2928 | 1.0 | 0.8987 |
| 2.6802 | 70.93 | 532 | 2.2408 | 1.0 | 0.8499 |
| 2.6802 | 72.0 | 540 | 2.2057 | 1.0 | 0.8463 |
| 2.6802 | 72.93 | 547 | 2.1440 | 1.0 | 0.8047 |
| 2.6802 | 74.0 | 555 | 2.1055 | 1.0 | 0.7975 |
| 2.6802 | 74.93 | 562 | 2.0576 | 1.0 | 0.7729 |
| 2.6802 | 76.0 | 570 | 2.0157 | 1.0 | 0.7631 |
| 2.6802 | 76.93 | 577 | 1.9685 | 1.0 | 0.7418 |
| 2.6802 | 78.0 | 585 | 1.9267 | 1.0 | 0.7197 |
| 2.6802 | 78.93 | 592 | 1.8942 | 1.0 | 0.7114 |
| 2.3153 | 80.0 | 600 | 1.8437 | 1.0 | 0.6593 |
| 2.3153 | 80.93 | 607 | 1.8056 | 1.0 | 0.6159 |
| 2.3153 | 82.0 | 615 | 1.7832 | 1.0 | 0.6221 |
| 2.3153 | 82.93 | 622 | 1.7551 | 0.9978 | 0.5917 |
| 2.3153 | 84.0 | 630 | 1.7235 | 0.9956 | 0.5548 |
| 2.3153 | 84.93 | 637 | 1.7026 | 0.9956 | 0.5476 |
| 2.3153 | 86.0 | 645 | 1.6728 | 0.9934 | 0.5107 |
| 2.3153 | 86.93 | 652 | 1.6532 | 0.9934 | 0.4904 |
| 2.3153 | 88.0 | 660 | 1.6387 | 0.9934 | 0.4828 |
| 2.3153 | 88.93 | 667 | 1.6284 | 0.9934 | 0.4763 |
| 2.3153 | 90.0 | 675 | 1.6146 | 0.9934 | 0.4615 |
| 2.3153 | 90.93 | 682 | 1.6049 | 0.9934 | 0.4514 |
| 2.3153 | 92.0 | 690 | 1.5985 | 0.9934 | 0.4470 |
| 2.3153 | 92.93 | 697 | 1.5960 | 0.9934 | 0.4459 |
| 2.0162 | 93.33 | 700 | 1.5957 | 0.9934 | 0.4456 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.0.1+cu117
- Datasets 2.14.4
- Tokenizers 0.13.3
