Hugging Face's logo

asr-africa
/
wav2vec2-pretrained-asr-africa-naijavoices-hausa-20hr-v1

Automatic Speech Recognition
Transformers
Safetensors
Hausa
wav2vec2
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Community
wav2vec2-pretrained-asr-africa-naijavoices-hausa-20hr-v1 / README.md
Alvin-Nahabwe's picture
Alvin-Nahabwe
End of training
7b7d58a verified
preview code
|
raw
history blame contribute delete
8.92 kB
metadata
library_name: transformers
language:
  - ha
base_model: Alvin-Nahabwe/wav2vec2-pretrained-asr-africa
tags:
  - generated_from_trainer
datasets:
  - naijavoices/naijavoices-dataset
metrics:
  - wer
model-index:
  - name: Wav2Vec2-ASR-Africa Hausa - Alvin Nahabwe
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: NaijaVoices
          type: naijavoices/naijavoices-dataset
        metrics:
          - name: Wer
            type: wer
            value: 0.4449253474157941

Wav2Vec2-ASR-Africa Hausa - Alvin Nahabwe

This model is a fine-tuned version of Alvin-Nahabwe/wav2vec2-pretrained-asr-africa on the NaijaVoices dataset. It achieves the following results on the evaluation set:

  • Cer: 0.1183
  • Loss: 1.4415
  • Wer: 0.4449

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: 0.0005
  • train_batch_size: 32
  • eval_batch_size: 16
  • seed: 42
  • optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.05
  • num_epochs: 100.0
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Cer Validation Loss Wer
2.8513 1.0 713 0.2910 0.8985 0.8985
0.9475 2.0 1426 0.2192 0.7238 0.7567
0.8134 3.0 2139 0.1993 0.6482 0.6990
0.7329 4.0 2852 0.1849 0.6005 0.6628
0.6738 5.0 3565 0.1738 0.5939 0.6399
0.6331 6.0 4278 0.1703 0.5654 0.6220
0.5909 7.0 4991 0.1593 0.5385 0.5846
0.5591 8.0 5704 0.1591 0.5353 0.5803
0.5277 9.0 6417 0.1560 0.5261 0.5691
0.5005 10.0 7130 0.1506 0.5118 0.5559
0.4744 11.0 7843 0.1488 0.5212 0.5463
0.4498 12.0 8556 0.1473 0.5317 0.5484
0.4229 13.0 9269 0.1441 0.5332 0.5348
0.4046 14.0 9982 0.1450 0.5304 0.5323
0.3846 15.0 10695 0.1438 0.5677 0.5283
0.3627 16.0 11408 0.1431 0.5575 0.5290
0.3419 17.0 12121 0.1428 0.5930 0.5268
0.3275 18.0 12834 0.1435 0.5773 0.5248
0.309 19.0 13547 0.1405 0.6326 0.5205
0.2974 20.0 14260 0.1404 0.6192 0.5173
0.283 21.0 14973 0.1425 0.6331 0.5238
0.2684 22.0 15686 0.1418 0.6447 0.5189
0.2578 23.0 16399 0.1395 0.7003 0.5125
0.2502 24.0 17112 0.1404 0.6843 0.5158
0.2392 25.0 17825 0.1372 0.7490 0.5061
0.2311 26.0 18538 0.1369 0.6989 0.5072
0.2223 27.0 19251 0.1382 0.7487 0.5063
0.2168 28.0 19964 0.1412 0.7233 0.5075
0.2122 29.0 20677 0.1350 0.8055 0.5023
0.2064 30.0 21390 0.1356 0.7682 0.5015
0.2002 31.0 22103 0.1369 0.8371 0.5025
0.1953 32.0 22816 0.1345 0.8060 0.4984
0.1903 33.0 23529 0.1356 0.8662 0.5025
0.1875 34.0 24242 0.1343 0.7856 0.4960
0.184 35.0 24955 0.1363 0.8334 0.4967
0.1792 36.0 25668 0.1368 0.8326 0.4984
0.1775 37.0 26381 0.1335 0.8292 0.4928
0.1742 38.0 27094 0.1352 0.8251 0.4962
0.1716 39.0 27807 0.1315 0.8758 0.4868
0.1673 40.0 28520 0.1303 0.8811 0.4864
0.1647 41.0 29233 0.1309 0.9360 0.4864
0.163 42.0 29946 0.1326 0.9481 0.4893
0.1615 43.0 30659 0.1337 0.9580 0.4924
0.1575 44.0 31372 0.1314 0.8831 0.4850
0.1556 45.0 32085 0.1316 0.9683 0.4830
0.1531 46.0 32798 0.1325 0.9607 0.4871
0.151 47.0 33511 0.1295 0.9514 0.4792
0.1504 48.0 34224 0.1285 0.9855 0.4785
0.1468 49.0 34937 0.1287 0.9842 0.4821
0.1454 50.0 35650 0.1282 0.9534 0.4782
0.1416 51.0 36363 0.1267 0.9950 0.4734
0.1407 52.0 37076 0.1274 0.9657 0.4754
0.1356 53.0 37789 0.1275 0.9976 0.4723
0.1349 54.0 38502 0.1275 1.0288 0.4740
0.1352 55.0 39215 0.1284 1.0283 0.4753
0.1334 56.0 39928 0.1274 1.0172 0.4742
0.1317 57.0 40641 0.1262 1.0150 0.4725
0.1281 58.0 41354 0.1269 1.0258 0.4705
0.1272 59.0 42067 0.1270 1.0485 0.4713
0.1261 60.0 42780 0.1257 1.0763 0.4674
0.1257 61.0 43493 0.1264 1.0779 0.4717
0.1223 62.0 44206 0.1267 1.0957 0.4697
0.1213 63.0 44919 0.1240 1.0922 0.4653
0.1191 64.0 45632 0.1252 1.0292 0.4675
0.1163 65.0 46345 0.1235 1.1721 0.4635
0.1152 66.0 47058 0.1237 1.0529 0.4625
0.1139 67.0 47771 0.1244 1.1248 0.4624
0.1122 68.0 48484 0.1239 1.1673 0.4615
0.1107 69.0 49197 0.1237 1.1475 0.4630
0.1097 70.0 49910 0.1238 1.1714 0.4639
0.1072 71.0 50623 0.1232 1.1740 0.4620
0.1059 72.0 51336 0.1229 1.2115 0.4613
0.104 73.0 52049 0.1230 1.1782 0.4640
0.101 74.0 52762 0.1225 1.2219 0.4621
0.1 75.0 53475 0.1209 1.2251 0.4564
0.1006 76.0 54188 0.1217 1.2656 0.4591
0.0993 77.0 54901 0.1215 1.2014 0.4600
0.0975 78.0 55614 0.1208 1.2231 0.4564
0.094 79.0 56327 0.1207 1.2751 0.4536
0.0963 80.0 57040 0.1205 1.2659 0.4531
0.0934 81.0 57753 0.1207 1.3305 0.4548
0.092 82.0 58466 0.1207 1.3029 0.4527
0.0905 83.0 59179 0.1210 1.2916 0.4554
0.0899 84.0 59892 0.1199 1.3628 0.4513
0.0876 85.0 60605 0.1200 1.3117 0.4507
0.086 86.0 61318 0.1193 1.3287 0.4491
0.0861 87.0 62031 0.1197 1.3370 0.4504
0.0844 88.0 62744 0.1197 1.3742 0.4492
0.0831 89.0 63457 0.1192 1.3519 0.4498
0.0824 90.0 64170 0.1191 1.3569 0.4493
0.0811 91.0 64883 0.1201 1.3567 0.4495
0.0798 92.0 65596 0.1189 1.3839 0.4484
0.0797 93.0 66309 0.1191 1.3982 0.4477
0.0786 94.0 67022 0.1190 1.4314 0.4499
0.0776 95.0 67735 0.1189 1.4345 0.4462
0.0773 96.0 68448 0.1185 1.4470 0.4452
0.0772 97.0 69161 0.1186 1.4376 0.4453
0.0754 98.0 69874 0.1183 1.4523 0.4453
0.0757 99.0 70587 0.1184 1.4372 0.4454
0.0751 100.0 71300 0.1183 1.4415 0.4449

Framework versions

  • Transformers 4.47.1
  • Pytorch 2.4.1+cu124
  • Datasets 3.2.0
  • Tokenizers 0.21.0