You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this model content.

wav2vec2-xlsr-luganda

This model is a fine-tuned version of sulaimank/wav2vec2-xlsr-CV_Fleurs_AMMI_ALFFA-swahili-200hrs on the None dataset. It achieves the following results on the evaluation set:

  • Cer: 0.0223
  • Loss: 0.2018
  • Wer: 0.1123

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: 4
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 16
  • optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: linear
  • num_epochs: 50
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Cer Validation Loss Wer
0.8458 1.0 16873 0.0601 0.2206 0.3081
0.4665 2.0 33746 0.0503 0.1822 0.2534
0.3881 3.0 50619 0.0442 0.1644 0.2235
0.3411 4.0 67492 0.0403 0.1545 0.2041
0.3080 5.0 84365 0.0378 0.1489 0.1922
0.2828 6.0 101238 0.0359 0.1406 0.1830
0.2626 7.0 118111 0.0346 0.1372 0.1770
0.2443 8.0 134984 0.0345 0.1338 0.1736
0.2290 9.0 151857 0.0324 0.1327 0.1651
0.2146 10.0 168730 0.0319 0.1308 0.1642
0.2027 11.0 185603 0.0310 0.1289 0.1600
0.1907 12.0 202476 0.0314 0.1315 0.1593
0.1800 13.0 219349 0.0303 0.1275 0.1557
0.1703 14.0 236222 0.0301 0.1323 0.1534
0.1614 15.0 253095 0.0294 0.1319 0.1510
0.1531 16.0 269968 0.0291 0.1333 0.1461
0.1453 17.0 286841 0.0280 0.1324 0.1438
0.1388 18.0 303714 0.0282 0.1350 0.1443
0.1318 19.0 320587 0.0280 0.1405 0.1425
0.1250 20.0 337460 0.0274 0.1419 0.1385
0.1200 21.0 354333 0.0271 0.1431 0.1377
0.1148 22.0 371206 0.0266 0.1485 0.1362
0.1094 23.0 388079 0.0262 0.1470 0.1338
0.1045 24.0 404952 0.0265 0.1534 0.1341
0.1010 25.0 421825 0.0258 0.1549 0.1305
0.0971 26.0 438698 0.0252 0.1574 0.1285
0.0926 27.0 455571 0.0256 0.1623 0.1306
0.0892 28.0 472444 0.0251 0.1620 0.1266
0.0862 29.0 489317 0.0249 0.1673 0.1253
0.0834 30.0 506190 0.0247 0.1671 0.1245
0.0800 31.0 523063 0.0245 0.1705 0.1237
0.0778 32.0 539936 0.0244 0.1767 0.1232
0.0757 33.0 556809 0.0244 0.1711 0.1217
0.0731 34.0 573682 0.0238 0.1778 0.1204
0.0705 35.0 590555 0.0234 0.1758 0.1188
0.0686 36.0 607428 0.0233 0.1754 0.1183
0.0662 37.0 624301 0.0234 0.1824 0.1172
0.0648 38.0 641174 0.0231 0.1818 0.1163
0.0634 39.0 658047 0.0232 0.1834 0.1174
0.0614 40.0 674920 0.0232 0.1844 0.1170
0.0603 41.0 691793 0.0230 0.1864 0.1165
0.0587 42.0 708666 0.0232 0.1926 0.1173
0.0573 43.0 725539 0.0232 0.1943 0.1175
0.0562 44.0 742412 0.0229 0.1976 0.1152
0.0551 45.0 759285 0.0227 0.1952 0.1142
0.0537 46.0 776158 0.0227 0.1963 0.1136
0.0532 47.0 793031 0.0225 0.2008 0.1138
0.0522 48.0 809904 0.0224 0.2005 0.1130
0.0520 49.0 826777 0.0224 0.2009 0.1126
0.0509 50.0 843650 0.0223 0.2018 0.1123

Framework versions

  • Transformers 5.4.0
  • Pytorch 2.11.0+cu130
  • Datasets 3.6.0
  • Tokenizers 0.22.2
Downloads last month
16
Safetensors
Model size
0.3B params
Tensor type
F32
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for sulaimank/wav2vec2-xlsr-luganda

Space using sulaimank/wav2vec2-xlsr-luganda 1