Edit model card

swbd-5percent-supervised

This model is a fine-tuned version of facebook/wav2vec2-large-lv60 on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.6970
  • Wer: 0.1352

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: 0.0001
  • train_batch_size: 8
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 1000
  • num_epochs: 30
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
6.8534 0.64 1000 2.9535 1.0
1.8605 1.28 2000 0.7878 0.3719
0.9862 1.92 3000 0.5906 0.2684
0.8405 2.56 4000 0.5555 0.2151
0.6972 3.2 5000 0.5905 0.1992
0.6033 3.84 6000 0.4867 0.1781
0.5393 4.48 7000 0.5447 0.1805
0.529 5.12 8000 0.5398 0.1746
0.5072 5.77 9000 0.5093 0.1706
0.4331 6.41 10000 0.4990 0.1627
0.4837 7.05 11000 0.5319 0.1634
0.3867 7.69 12000 0.4866 0.1595
0.345 8.33 13000 0.5202 0.1582
0.372 8.97 14000 0.5396 0.1547
0.355 9.61 15000 0.5992 0.1493
0.3258 10.25 16000 0.5247 0.1527
0.3327 10.89 17000 0.5664 0.1512
0.3422 11.53 18000 0.5819 0.1456
0.2815 12.17 19000 0.5692 0.1453
0.2719 12.81 20000 0.5012 0.1476
0.2838 13.45 21000 0.5286 0.1454
0.2418 14.09 22000 0.6238 0.1486
0.2412 14.73 23000 0.5889 0.1456
0.2227 15.37 24000 0.5901 0.1459
0.2129 16.02 25000 0.5959 0.1454
0.2071 16.66 26000 0.6259 0.1427
0.2185 17.3 27000 0.6581 0.1437
0.1982 17.94 28000 0.6194 0.1411
0.1928 18.58 29000 0.5940 0.1409
0.1885 19.22 30000 0.6733 0.1417
0.1835 19.86 31000 0.6363 0.1393
0.1756 20.5 32000 0.6675 0.1382
0.1776 21.14 33000 0.6147 0.1407
0.1758 21.78 34000 0.6405 0.1420
0.1645 22.42 35000 0.6999 0.1401
0.1631 23.06 36000 0.6224 0.1385
0.1494 23.7 37000 0.6639 0.1374
0.1472 24.34 38000 0.6471 0.1373
0.1514 24.98 39000 0.6570 0.1395
0.1527 25.62 40000 0.6876 0.1375
0.1514 26.27 41000 0.6835 0.1376
0.1344 26.91 42000 0.6987 0.1372
0.1267 27.55 43000 0.7026 0.1362
0.1384 28.19 44000 0.7021 0.1366
0.1264 28.83 45000 0.7016 0.1355
0.1227 29.47 46000 0.6970 0.1352

Framework versions

  • Transformers 4.14.1
  • Pytorch 1.10.2
  • Datasets 1.18.2
  • Tokenizers 0.10.3
Downloads last month
13