Modelpublic

alinerodrigues/wav2vec2-large-xlsr-facebook-300m-aug-texts-exp-1

sourceHugging Faceupdated 1y agoView on Hugging Face

0likes7downloads

Model Card

wav2vec2-large-xlsr-facebook-300m-aug-texts-exp-1

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

—Loss: 0.2660
—Wer: 0.1910
—Cer: 0.1174
—Per: 0.1355

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
—trainbatchsize: 16
—evalbatchsize: 8
—seed: 42
—gradientaccumulationsteps: 2
—totaltrainbatch_size: 32
—optimizer: Use OptimizerNames.ADAMWTORCHFUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
—lrschedulertype: linear
—num_epochs: 30
—mixedprecisiontraining: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer	Per
0.4757	1.0	64	0.3030	0.1982	0.1230	0.1417
0.203	2.0	128	0.2749	0.1926	0.1183	0.1366
0.203	3.0	192	0.2866	0.1897	0.1177	0.1367
0.1566	4.0	256	0.2787	0.1876	0.1178	0.1362
0.1269	5.0	320	0.2748	0.1968	0.1193	0.1378
0.1269	6.0	384	0.2660	0.1910	0.1173	0.1361
0.1302	7.0	448	0.2984	0.1882	0.1189	0.1373
0.1191	8.0	512	0.2732	0.1902	0.1180	0.1368
0.1191	9.0	576	0.3016	0.1894	0.1193	0.1379
0.1132	10.0	640	0.2839	0.1888	0.1183	0.1374
0.1063	11.0	704	0.2803	0.1897	0.1173	0.1349
0.1063	12.0	768	0.2825	0.1868	0.1178	0.1363
0.1053	13.0	832	0.2684	0.1906	0.1184	0.1374
0.1053	14.0	896	0.2836	0.1904	0.1182	0.1365
0.0901	15.0	960	0.2830	0.1839	0.1164	0.1347
0.0938	16.0	1024	0.3050	0.1887	0.1184	0.1372

Framework versions

—Transformers 4.55.2
—Pytorch 2.8.0+cu126
—Datasets 4.0.0
—Tokenizers 0.21.4