Hugging Face's logo

sulaimank
/
hubert-large-ls960-ft-lg-CV-v1

Automatic Speech Recognition
Transformers
Safetensors
hubert
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Community
hubert-large-ls960-ft-lg-CV-v1 / README.md
sulaimank's picture
sulaimank
End of training
04e3399 verified
preview code
|
raw
history blame contribute delete
9.01 kB
metadata
library_name: transformers
license: apache-2.0
base_model: facebook/hubert-large-ls960-ft
tags:
  - generated_from_trainer
datasets:
  - common_voice_17_0
metrics:
  - wer
model-index:
  - name: hubert-large-ls960-ft-lg-CV-v1
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: common_voice_17_0
          type: common_voice_17_0
          config: lg
          split: None
          args: lg
        metrics:
          - name: Wer
            type: wer
            value: 0.20413735167489236

hubert-large-ls960-ft-lg-CV-v1

This model is a fine-tuned version of facebook/hubert-large-ls960-ft on the common_voice_17_0 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.6251
  • Wer: 0.2041
  • Cer: 0.0609

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: 0.0003
  • train_batch_size: 8
  • eval_batch_size: 4
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 16
  • optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: linear
  • num_epochs: 100
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer Cer
0.5741 1.0 4442 0.4573 0.4271 0.1144
0.3189 2.0 8884 0.3821 0.3444 0.0932
0.2692 3.0 13326 0.3881 0.3310 0.0912
0.2405 4.0 17768 0.3453 0.3136 0.0854
0.2191 5.0 22210 0.3476 0.2931 0.0823
0.2039 6.0 26652 0.3841 0.2880 0.0825
0.1913 7.0 31094 0.3532 0.2869 0.0798
0.18 8.0 35536 0.3727 0.2849 0.0823
0.1708 9.0 39978 0.3410 0.2773 0.0785
0.1624 10.0 44420 0.3604 0.2705 0.0794
0.1552 11.0 48862 0.3589 0.2661 0.0765
0.1485 12.0 53304 0.3614 0.2687 0.0770
0.1418 13.0 57746 0.3500 0.2637 0.0762
0.1358 14.0 62188 0.3713 0.2628 0.0766
0.131 15.0 66630 0.3908 0.2603 0.0758
0.1255 16.0 71072 0.4089 0.2608 0.0758
0.1205 17.0 75514 0.3848 0.2595 0.0742
0.1162 18.0 79956 0.3554 0.2594 0.0739
0.1125 19.0 84398 0.3461 0.2593 0.0742
0.1073 20.0 88840 0.3663 0.2545 0.0729
0.1039 21.0 93282 0.4556 0.2578 0.0743
0.1 22.0 97724 0.4258 0.2504 0.0724
0.0965 23.0 102166 0.4246 0.2545 0.0754
0.0931 24.0 106608 0.4570 0.2603 0.0757
0.0894 25.0 111050 0.4039 0.2488 0.0732
0.0865 26.0 115492 0.4119 0.2510 0.0720
0.083 27.0 119934 0.4227 0.2454 0.0716
0.0805 28.0 124376 0.4424 0.2541 0.0728
0.0777 29.0 128818 0.4061 0.2457 0.0709
0.0762 30.0 133260 0.4114 0.2450 0.0704
0.0724 31.0 137702 0.4599 0.2516 0.0719
0.0711 32.0 142144 0.4311 0.2466 0.0714
0.069 33.0 146586 0.4517 0.2482 0.0717
0.0673 34.0 151028 0.4728 0.2467 0.0712
0.0655 35.0 155470 0.4542 0.2437 0.0713
0.0634 36.0 159912 0.4546 0.2480 0.0713
0.0612 37.0 164354 0.4852 0.2479 0.0718
0.0607 38.0 168796 0.4892 0.2433 0.0705
0.0585 39.0 173238 0.4686 0.2416 0.0702
0.0573 40.0 177680 0.4725 0.2412 0.0710
0.0556 41.0 182122 0.4737 0.2385 0.0696
0.0548 42.0 186564 0.4964 0.2448 0.0704
0.0527 43.0 191006 0.5236 0.2429 0.0706
0.052 44.0 195448 0.5130 0.2415 0.0714
0.0503 45.0 199890 0.4936 0.2375 0.0688
0.0496 46.0 204332 0.5120 0.2336 0.0680
0.048 47.0 208774 0.4964 0.2362 0.0694
0.0473 48.0 213216 0.5200 0.2372 0.0687
0.0465 49.0 217658 0.5433 0.2424 0.0708
0.0447 50.0 222100 0.5008 0.2335 0.0680
0.0444 51.0 226542 0.5024 0.2247 0.0668
0.0431 52.0 230984 0.5003 0.2307 0.0669
0.0423 53.0 235426 0.4892 0.2331 0.0676
0.0403 54.0 239868 0.5495 0.2316 0.0679
0.0406 55.0 244310 0.5193 0.2278 0.0661
0.0391 56.0 248752 0.5961 0.2331 0.0687
0.0389 57.0 253194 0.5227 0.2297 0.0667
0.0379 58.0 257636 0.5506 0.2295 0.0672
0.0366 59.0 262078 0.5725 0.2231 0.0673
0.0357 60.0 266520 0.5493 0.2280 0.0662
0.0357 61.0 270962 0.5355 0.2269 0.0656
0.035 62.0 275404 0.5430 0.2226 0.0653
0.0343 63.0 279846 0.5375 0.2211 0.0644
0.0334 64.0 284288 0.5769 0.2248 0.0668
0.0333 65.0 288730 0.5763 0.2183 0.0642
0.0322 66.0 293172 0.5787 0.2190 0.0653
0.0314 67.0 297614 0.5564 0.2207 0.0642
0.0305 68.0 302056 0.5813 0.2208 0.0666
0.03 69.0 306498 0.5837 0.2217 0.0647
0.0292 70.0 310940 0.5723 0.2238 0.0649
0.0284 71.0 315382 0.5503 0.2218 0.0645
0.0285 72.0 319824 0.5615 0.2187 0.0636
0.0276 73.0 324266 0.5725 0.2178 0.0650
0.0273 74.0 328708 0.5483 0.2187 0.0634
0.027 75.0 333150 0.5627 0.2148 0.0632
0.026 76.0 337592 0.5610 0.2203 0.0655
0.0253 77.0 342034 0.5776 0.2153 0.0635
0.0248 78.0 346476 0.5823 0.2173 0.0643
0.0242 79.0 350918 0.5968 0.2172 0.0639
0.0241 80.0 355360 0.6121 0.2185 0.0647
0.0232 81.0 359802 0.5909 0.2140 0.0648
0.0227 82.0 364244 0.6262 0.2209 0.0663
0.0224 83.0 368686 0.5913 0.2137 0.0645
0.0215 84.0 373128 0.6057 0.2141 0.0642
0.0212 85.0 377570 0.6079 0.2135 0.0635
0.0209 86.0 382012 0.6067 0.2117 0.0639
0.0201 87.0 386454 0.6119 0.2108 0.0638
0.0199 88.0 390896 0.6298 0.2112 0.0638
0.0194 89.0 395338 0.6054 0.2083 0.0620
0.0192 90.0 399780 0.6238 0.2083 0.0634
0.0184 91.0 404222 0.6293 0.2099 0.0630
0.0184 92.0 408664 0.6166 0.2058 0.0611
0.0182 93.0 413106 0.6175 0.2072 0.0618
0.0179 94.0 417548 0.6196 0.2061 0.0610
0.0176 95.0 421990 0.6181 0.2059 0.0614
0.0174 96.0 426432 0.6187 0.2039 0.0606
0.0167 97.0 430874 0.6381 0.2064 0.0615
0.017 98.0 435316 0.6268 0.2049 0.0611
0.0165 99.0 439758 0.6262 0.2041 0.0610
0.0166 100.0 444200 0.6251 0.2041 0.0609

Framework versions

  • Transformers 4.46.3
  • Pytorch 2.1.0+cu118
  • Datasets 3.1.0
  • Tokenizers 0.20.3