Hugging Face's logo

HarrisDePerceptron
/
xls-r-300m-ur-cv7

Automatic Speech Recognition
Transformers
PyTorch
Urdu
wav2vec2
mozilla-foundation/common_voice_7_0
Generated from Trainer
Inference Endpoints
Model card Files Community
xls-r-300m-ur-cv7 / README.md
HarrisDePerceptron's picture
HarrisDePerceptron
update model card README.md
b7c3311
preview code
|
raw
history blame
8.85 kB
metadata
license: apache-2.0
tags:
  - generated_from_trainer
datasets:
  - common_voice
model-index:
  - name: ''
    results: []

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice dataset. It achieves the following results on the evaluation set:

  • Loss: 1.7564
  • Wer: 0.7201

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: 7.5e-05
  • train_batch_size: 8
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 16
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 1000
  • num_epochs: 500.0
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
18.5133 4.17 100 12.0376 1.0
7.9152 8.33 200 6.2406 1.0
5.363 12.5 300 4.6649 1.0
4.05 16.67 400 3.5798 1.0
3.3467 20.83 500 3.1930 1.0
3.1638 25.0 600 3.0984 1.0
3.043 29.17 700 2.9707 0.9983
2.9566 33.33 800 2.9296 1.0
2.8994 37.5 900 2.8290 0.9983
2.6469 41.67 1000 2.2443 0.9765
2.1557 45.83 1100 1.5079 0.8204
1.7524 50.0 1200 1.2472 0.7594
1.54 54.17 1300 1.1612 0.7097
1.3985 58.33 1400 1.1158 0.7358
1.2869 62.5 1500 1.0454 0.6931
1.1952 66.67 1600 1.0130 0.6853
1.1022 70.83 1700 1.0176 0.6966
1.0346 75.0 1800 1.0053 0.6818
0.9707 79.17 1900 1.0224 0.6713
0.917 83.33 2000 1.0383 0.6530
0.8574 87.5 2100 1.0632 0.6757
0.8021 91.67 2200 1.0126 0.6443
0.7563 95.83 2300 1.0678 0.6713
0.709 100.0 2400 1.0756 0.6757
0.6775 104.17 2500 1.0397 0.6914
0.6325 108.33 2600 1.0556 0.6513
0.617 112.5 2700 1.1219 0.6922
0.5801 116.67 2800 1.1125 0.6870
0.5367 120.83 2900 1.1397 0.6565
0.5132 125.0 3000 1.1678 0.7001
0.4948 129.17 3100 1.1276 0.6643
0.457 133.33 3200 1.1465 0.6792
0.4538 137.5 3300 1.1435 0.6809
0.4227 141.67 3400 1.1457 0.6600
0.4083 145.83 3500 1.1793 0.6652
0.3965 150.0 3600 1.2435 0.6879
0.382 154.17 3700 1.2229 0.6827
0.3452 158.33 3800 1.2297 0.6879
0.3434 162.5 3900 1.2350 0.6888
0.3276 166.67 4000 1.2186 0.6922
0.3052 170.83 4100 1.2338 0.6870
0.3025 175.0 4200 1.2724 0.7079
0.2916 179.17 4300 1.2758 0.6975
0.2709 183.33 4400 1.2726 0.6748
0.2707 187.5 4500 1.2643 0.6957
0.262 191.67 4600 1.3214 0.7132
0.2453 195.83 4700 1.2953 0.6861
0.248 200.0 4800 1.3622 0.6774
0.2325 204.17 4900 1.3594 0.6835
0.2124 208.33 5000 1.3367 0.6652
0.2253 212.5 5100 1.4157 0.6879
0.2059 216.67 5200 1.4360 0.7132
0.1951 220.83 5300 1.4606 0.7158
0.1861 225.0 5400 1.4695 0.7018
0.1916 229.17 5500 1.4031 0.6739
0.1822 233.33 5600 1.4426 0.6870
0.1684 237.5 5700 1.4069 0.7053
0.1719 241.67 5800 1.4766 0.6966
0.1569 245.83 5900 1.4509 0.6931
0.159 250.0 6000 1.4467 0.7097
0.1476 254.17 6100 1.4617 0.6870
0.1497 258.33 6200 1.4460 0.6844
0.1446 262.5 6300 1.5557 0.7088
0.1389 266.67 6400 1.4886 0.7140
0.1331 270.83 6500 1.5526 0.7062
0.1344 275.0 6600 1.5419 0.7027
0.1198 279.17 6700 1.5641 0.7001
0.1242 283.33 6800 1.5390 0.7062
0.12 287.5 6900 1.5406 0.7105
0.1096 291.67 7000 1.5737 0.6975
0.113 295.83 7100 1.5495 0.7210
0.108 300.0 7200 1.5375 0.6949
0.1072 304.17 7300 1.5337 0.7010
0.0979 308.33 7400 1.5927 0.7062
0.0983 312.5 7500 1.5882 0.6844
0.0977 316.67 7600 1.6189 0.6957
0.0947 320.83 7700 1.5098 0.6818
0.0996 325.0 7800 1.6269 0.7254
0.0846 329.17 7900 1.6367 0.7088
0.0953 333.33 8000 1.5965 0.7123
0.0906 337.5 8100 1.6096 0.7123
0.093 341.67 8200 1.5953 0.6983
0.0784 345.83 8300 1.5884 0.6914
0.0769 350.0 8400 1.5794 0.6870
0.0782 354.17 8500 1.6581 0.6818
0.0764 358.33 8600 1.6555 0.7088
0.073 362.5 8700 1.6466 0.6931
0.0703 366.67 8800 1.6615 0.7114
0.0707 370.83 8900 1.6743 0.7079
0.0647 375.0 9000 1.6452 0.7167
0.0614 379.17 9100 1.7082 0.7123
0.0646 383.33 9200 1.6848 0.7184
0.0648 387.5 9300 1.6581 0.7088
0.0625 391.67 9400 1.7315 0.7341
0.0637 395.83 9500 1.6831 0.7027
0.0558 400.0 9600 1.7159 0.7280
0.0563 404.17 9700 1.7475 0.7158
0.0568 408.33 9800 1.6776 0.6992
0.0574 412.5 9900 1.7150 0.6983
0.0561 416.67 10000 1.7315 0.7140
0.0494 420.83 10100 1.6869 0.7219
0.0495 425.0 10200 1.7500 0.7262
0.0542 429.17 10300 1.7298 0.7271
0.0509 433.33 10400 1.7334 0.7262
0.046 437.5 10500 1.7048 0.7193
0.0423 441.67 10600 1.7168 0.7193
0.0477 445.83 10700 1.7388 0.7210
0.0436 450.0 10800 1.7279 0.7167
0.0466 454.17 10900 1.6968 0.7053
0.0424 458.33 11000 1.7237 0.7184
0.0447 462.5 11100 1.7218 0.7184
0.0455 466.67 11200 1.7506 0.7219
0.0446 470.83 11300 1.7542 0.7280
0.043 475.0 11400 1.7501 0.7201
0.0397 479.17 11500 1.7837 0.7245
0.0402 483.33 11600 1.7762 0.7175
0.039 487.5 11700 1.7771 0.7262
0.0402 491.67 11800 1.7564 0.7219
0.0368 495.83 11900 1.7553 0.7193
0.0395 500.0 12000 1.7564 0.7201

Framework versions

  • Transformers 4.17.0.dev0
  • Pytorch 1.10.2+cu102
  • Datasets 1.18.3
  • Tokenizers 0.11.0