|
--- |
|
language: |
|
- en |
|
license: mit |
|
base_model: facebook/w2v-bert-2.0 |
|
tags: |
|
- automatic-speech-recognition |
|
- mozilla-foundation/common_voice_16_0 |
|
- generated_from_trainer |
|
metrics: |
|
- wer |
|
model-index: |
|
- name: wav2vec2-bert-CV16-en-cv-2 |
|
results: [] |
|
--- |
|
|
|
<!-- 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-bert-CV16-en-cv-2 |
|
|
|
This model is a fine-tuned version of [facebook/w2v-bert-2.0](https://huggingface.co/facebook/w2v-bert-2.0) on the MOZILLA-FOUNDATION/COMMON_VOICE_16_0 - EN dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.4363 |
|
- Wer: 0.1304 |
|
- Cer: 0.0527 |
|
|
|
## 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: 12 |
|
- eval_batch_size: 12 |
|
- seed: 42 |
|
- distributed_type: multi-GPU |
|
- num_devices: 3 |
|
- gradient_accumulation_steps: 3 |
|
- total_train_batch_size: 108 |
|
- total_eval_batch_size: 36 |
|
- optimizer: Adam with betas=(0.9,0.96) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- lr_scheduler_warmup_steps: 10000 |
|
- num_epochs: 50.0 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer | |
|
|:-------------:|:-----:|:------:|:---------------:|:------:|:------:| |
|
| 0.1884 | 0.49 | 5000 | 0.3214 | 0.1872 | 0.0699 | |
|
| 0.1862 | 0.98 | 10000 | 0.2829 | 0.1759 | 0.0674 | |
|
| 0.1493 | 1.47 | 15000 | 0.2955 | 0.1680 | 0.0658 | |
|
| 0.1346 | 1.96 | 20000 | 0.2727 | 0.1649 | 0.0647 | |
|
| 0.1273 | 2.45 | 25000 | 0.2677 | 0.1608 | 0.0631 | |
|
| 0.1241 | 2.94 | 30000 | 0.2613 | 0.1594 | 0.0631 | |
|
| 0.1059 | 3.43 | 35000 | 0.2737 | 0.1600 | 0.0636 | |
|
| 0.1201 | 3.92 | 40000 | 0.2633 | 0.1553 | 0.0618 | |
|
| 0.0961 | 4.4 | 45000 | 0.2666 | 0.1549 | 0.0617 | |
|
| 0.0971 | 4.89 | 50000 | 0.2647 | 0.1536 | 0.0616 | |
|
| 0.088 | 5.38 | 55000 | 0.2504 | 0.1500 | 0.0597 | |
|
| 0.0931 | 5.87 | 60000 | 0.2494 | 0.1500 | 0.0599 | |
|
| 0.0906 | 6.36 | 65000 | 0.2604 | 0.1507 | 0.0602 | |
|
| 0.0788 | 6.85 | 70000 | 0.2629 | 0.1506 | 0.0603 | |
|
| 0.0692 | 7.34 | 75000 | 0.2500 | 0.1484 | 0.0590 | |
|
| 0.0896 | 7.83 | 80000 | 0.2525 | 0.1468 | 0.0586 | |
|
| 0.0794 | 8.32 | 85000 | 0.2494 | 0.1458 | 0.0583 | |
|
| 0.0908 | 8.81 | 90000 | 0.2539 | 0.1475 | 0.0587 | |
|
| 0.0646 | 9.3 | 95000 | 0.2539 | 0.1457 | 0.0578 | |
|
| 0.0877 | 9.79 | 100000 | 0.2503 | 0.1453 | 0.0583 | |
|
| 0.0657 | 10.28 | 105000 | 0.2589 | 0.1457 | 0.0586 | |
|
| 0.0715 | 10.77 | 110000 | 0.2638 | 0.1462 | 0.0586 | |
|
| 0.0704 | 11.26 | 115000 | 0.2562 | 0.1419 | 0.0571 | |
|
| 0.0582 | 11.75 | 120000 | 0.2663 | 0.1425 | 0.0577 | |
|
| 0.0583 | 12.23 | 125000 | 0.2615 | 0.1408 | 0.0573 | |
|
| 0.0776 | 12.72 | 130000 | 0.2593 | 0.1434 | 0.0575 | |
|
| 0.0624 | 13.21 | 135000 | 0.2603 | 0.1438 | 0.0577 | |
|
| 0.0619 | 13.7 | 140000 | 0.2512 | 0.1391 | 0.0564 | |
|
| 0.0642 | 14.19 | 145000 | 0.2539 | 0.1386 | 0.0558 | |
|
| 0.0593 | 14.68 | 150000 | 0.2609 | 0.1406 | 0.0567 | |
|
| 0.0596 | 15.17 | 155000 | 0.2567 | 0.1390 | 0.0562 | |
|
| 0.0671 | 15.66 | 160000 | 0.2618 | 0.1404 | 0.0574 | |
|
| 0.0537 | 16.15 | 165000 | 0.2668 | 0.1391 | 0.0565 | |
|
| 0.0543 | 16.64 | 170000 | 0.2583 | 0.1379 | 0.0560 | |
|
| 0.056 | 17.13 | 175000 | 0.2612 | 0.1395 | 0.0564 | |
|
| 0.0605 | 17.62 | 180000 | 0.2654 | 0.1375 | 0.0557 | |
|
| 0.0536 | 18.11 | 185000 | 0.2703 | 0.1356 | 0.0549 | |
|
| 0.0469 | 18.6 | 190000 | 0.2571 | 0.1348 | 0.0548 | |
|
| 0.0548 | 19.09 | 195000 | 0.2621 | 0.1359 | 0.0551 | |
|
| 0.0507 | 19.58 | 200000 | 0.2628 | 0.1348 | 0.0549 | |
|
| 0.0513 | 20.06 | 205000 | 0.2722 | 0.1367 | 0.0558 | |
|
| 0.0423 | 20.55 | 210000 | 0.2752 | 0.1349 | 0.0546 | |
|
| 0.049 | 21.04 | 215000 | 0.2777 | 0.1358 | 0.0552 | |
|
| 0.0434 | 21.53 | 220000 | 0.2671 | 0.1336 | 0.0549 | |
|
| 0.0443 | 22.02 | 225000 | 0.2815 | 0.1333 | 0.0544 | |
|
| 0.0533 | 22.51 | 230000 | 0.2674 | 0.1334 | 0.0542 | |
|
| 0.0458 | 23.0 | 235000 | 0.2746 | 0.1320 | 0.0541 | |
|
| 0.0527 | 23.49 | 240000 | 0.2750 | 0.1351 | 0.0546 | |
|
| 0.0458 | 23.98 | 245000 | 0.2748 | 0.1322 | 0.0539 | |
|
| 0.0434 | 24.47 | 250000 | 0.2774 | 0.1317 | 0.0538 | |
|
| 0.0434 | 24.96 | 255000 | 0.2756 | 0.1322 | 0.0534 | |
|
| 0.041 | 25.45 | 260000 | 0.2786 | 0.1337 | 0.0542 | |
|
| 0.0408 | 25.94 | 265000 | 0.2785 | 0.1320 | 0.0534 | |
|
| 0.0486 | 26.43 | 270000 | 0.2882 | 0.1325 | 0.0536 | |
|
| 0.0469 | 26.92 | 275000 | 0.2796 | 0.1315 | 0.0532 | |
|
| 0.041 | 27.41 | 280000 | 0.2786 | 0.1319 | 0.0536 | |
|
| 0.0333 | 27.89 | 285000 | 0.2893 | 0.1316 | 0.0532 | |
|
| 0.0391 | 28.38 | 290000 | 0.2893 | 0.1318 | 0.0537 | |
|
| 0.0427 | 28.87 | 295000 | 0.3006 | 0.1326 | 0.0535 | |
|
| 0.0469 | 29.36 | 300000 | 0.2846 | 0.1308 | 0.0530 | |
|
| 0.0317 | 29.85 | 305000 | 0.3140 | 0.1311 | 0.0534 | |
|
| 0.0373 | 30.34 | 310000 | 0.2951 | 0.1314 | 0.0533 | |
|
| 0.0367 | 30.83 | 315000 | 0.2976 | 0.1309 | 0.0532 | |
|
| 0.0385 | 31.32 | 320000 | 0.3068 | 0.1311 | 0.0534 | |
|
| 0.0372 | 31.81 | 325000 | 0.3085 | 0.1295 | 0.0527 | |
|
| 0.0342 | 32.3 | 330000 | 0.3150 | 0.1290 | 0.0527 | |
|
| 0.035 | 32.79 | 335000 | 0.3133 | 0.1299 | 0.0530 | |
|
| 0.0331 | 33.28 | 340000 | 0.3201 | 0.1303 | 0.0530 | |
|
| 0.0334 | 33.77 | 345000 | 0.3310 | 0.1294 | 0.0527 | |
|
| 0.0353 | 34.26 | 350000 | 0.3105 | 0.1287 | 0.0523 | |
|
| 0.0367 | 34.75 | 355000 | 0.3180 | 0.1302 | 0.0530 | |
|
| 0.0397 | 35.24 | 360000 | 0.3322 | 0.1297 | 0.0528 | |
|
| 0.0384 | 35.72 | 365000 | 0.3290 | 0.1303 | 0.0530 | |
|
| 0.0349 | 36.21 | 370000 | 0.3358 | 0.1298 | 0.0529 | |
|
| 0.0352 | 36.7 | 375000 | 0.3351 | 0.1347 | 0.0546 | |
|
| 0.0333 | 37.19 | 380000 | 0.3420 | 0.1296 | 0.0528 | |
|
| 0.0282 | 37.68 | 385000 | 0.3426 | 0.1317 | 0.0534 | |
|
| 0.0247 | 38.17 | 390000 | 0.3606 | 0.1318 | 0.0531 | |
|
| 0.0312 | 38.66 | 395000 | 0.3509 | 0.1494 | 0.0601 | |
|
| 0.0288 | 39.15 | 400000 | 0.3516 | 0.1325 | 0.0536 | |
|
| 0.0281 | 39.64 | 405000 | 0.3489 | 0.1303 | 0.0528 | |
|
| 0.0208 | 40.13 | 410000 | 0.3661 | 0.1430 | 0.0576 | |
|
| 0.0276 | 40.62 | 415000 | 0.3620 | 0.1300 | 0.0524 | |
|
| 0.0253 | 41.11 | 420000 | 0.3786 | 0.1328 | 0.0538 | |
|
| 0.025 | 41.6 | 425000 | 0.3782 | 0.1321 | 0.0535 | |
|
| 0.02 | 42.09 | 430000 | 0.3721 | 0.1297 | 0.0523 | |
|
| 0.0192 | 42.58 | 435000 | 0.4099 | 0.1308 | 0.0528 | |
|
| 0.0197 | 43.07 | 440000 | 0.3970 | 0.1292 | 0.0525 | |
|
| 0.0177 | 43.55 | 445000 | 0.3946 | 0.1306 | 0.0531 | |
|
| 0.0185 | 44.04 | 450000 | 0.4060 | 0.1293 | 0.0526 | |
|
| 0.0176 | 44.53 | 455000 | 0.3968 | 0.1303 | 0.0529 | |
|
| 0.0126 | 45.02 | 460000 | 0.3994 | 0.1304 | 0.0529 | |
|
| 0.0142 | 45.51 | 465000 | 0.3975 | 0.1300 | 0.0527 | |
|
| 0.0122 | 46.0 | 470000 | 0.4055 | 0.1287 | 0.0523 | |
|
| 0.0115 | 46.49 | 475000 | 0.4211 | 0.1303 | 0.0526 | |
|
| 0.0102 | 46.98 | 480000 | 0.4148 | 0.1298 | 0.0525 | |
|
| 0.0119 | 47.47 | 485000 | 0.4238 | 0.1301 | 0.0527 | |
|
| 0.0098 | 47.96 | 490000 | 0.4293 | 0.1299 | 0.0526 | |
|
| 0.0125 | 48.45 | 495000 | 0.4375 | 0.1300 | 0.0526 | |
|
| 0.0095 | 48.94 | 500000 | 0.4268 | 0.1303 | 0.0527 | |
|
| 0.0055 | 49.43 | 505000 | 0.4286 | 0.1305 | 0.0527 | |
|
| 0.0089 | 49.92 | 510000 | 0.4371 | 0.1304 | 0.0528 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.37.0.dev0 |
|
- Pytorch 2.1.0+cu121 |
|
- Datasets 2.16.1 |
|
- Tokenizers 0.15.0 |
|
|
