Hugging Face's logo

ismichel
/
whisper-tiny-finetune

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
whisper
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
whisper-tiny-finetune / README.md
ismichel's picture
ismichel
End of training
b9fbb55 verified
preview code
|
raw
history blame contribute delete
7.66 kB
metadata
license: apache-2.0
base_model: openai/whisper-tiny.en
tags:
  - generated_from_trainer
metrics:
  - wer
model-index:
  - name: whisper-tiny-finetune
    results: []

whisper-tiny-finetune

This model is a fine-tuned version of openai/whisper-tiny.en on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.5924
  • Wer: 20.5257

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: 1e-05
  • train_batch_size: 128
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 500
  • training_steps: 1000

Training results

Training Loss Epoch Step Validation Loss Wer
3.9976 0.2778 10 3.9664 44.1176
3.9022 0.5556 20 3.8748 43.6170
3.7582 0.8333 30 3.7266 43.0538
3.5744 1.1111 40 3.5294 39.1427
3.2991 1.3889 50 3.2828 35.3567
3.0818 1.6667 60 2.9793 34.7935
2.6849 1.9444 70 2.5935 37.0150
2.2809 2.2222 80 2.0883 37.7034
1.6882 2.5 90 1.5027 38.6733
1.2127 2.7778 100 1.0275 39.0488
0.8696 3.0556 110 0.7946 29.4431
0.7336 3.3333 120 0.7152 28.1915
0.6793 3.6111 130 0.6666 27.2215
0.6489 3.8889 140 0.6346 26.0638
0.6353 4.1667 150 0.6115 24.8748
0.583 4.4444 160 0.5928 24.7497
0.5455 4.7222 170 0.5775 23.8110
0.487 5.0 180 0.5647 23.3417
0.4925 5.2778 190 0.5541 22.7159
0.4952 5.5556 200 0.5444 22.6533
0.4481 5.8333 210 0.5359 22.2153
0.4827 6.1111 220 0.5263 22.4030
0.3897 6.3889 230 0.5196 21.8085
0.3834 6.6667 240 0.5121 21.8711
0.3906 6.9444 250 0.5073 21.2140
0.3705 7.2222 260 0.5055 21.3705
0.3518 7.5 270 0.4980 21.2140
0.354 7.7778 280 0.4934 20.7447
0.3202 8.0556 290 0.4914 20.4318
0.2997 8.3333 300 0.4859 20.0563
0.2699 8.6111 310 0.4852 26.9399
0.2724 8.8889 320 0.4809 27.0338
0.2844 9.1667 330 0.4802 26.4393
0.2332 9.4444 340 0.4801 24.6558
0.2337 9.7222 350 0.4810 20.2753
0.2542 10.0 360 0.4731 20.5882
0.1986 10.2778 370 0.4779 20.1189
0.2023 10.5556 380 0.4767 24.6558
0.1864 10.8333 390 0.4763 20.3379
0.1873 11.1111 400 0.4765 20.6195
0.1595 11.3889 410 0.4831 20.4631
0.1581 11.6667 420 0.4872 20.2128
0.1663 11.9444 430 0.4851 20.0563
0.1282 12.2222 440 0.4864 19.9625
0.1138 12.5 450 0.4918 19.9937
0.1283 12.7778 460 0.4931 19.9312
0.0847 13.0556 470 0.4891 20.4944
0.0902 13.3333 480 0.5027 19.8999
0.0719 13.6111 490 0.5056 20.6821
0.1011 13.8889 500 0.5023 19.9937
0.0676 14.1667 510 0.5113 20.4005
0.0632 14.4444 520 0.5154 24.7184
0.0643 14.7222 530 0.5207 20.1502
0.053 15.0 540 0.5184 20.2753
0.0389 15.2778 550 0.5295 20.4631
0.0467 15.5556 560 0.5286 20.3066
0.0414 15.8333 570 0.5403 20.2753
0.0334 16.1111 580 0.5334 20.0876
0.0283 16.3889 590 0.5514 20.2441
0.0282 16.6667 600 0.5415 20.1815
0.0267 16.9444 610 0.5451 20.7447
0.019 17.2222 620 0.5483 20.3379
0.0202 17.5 630 0.5551 19.9625
0.0179 17.7778 640 0.5574 20.3066
0.0186 18.0556 650 0.5621 20.6821
0.0123 18.3333 660 0.5634 20.6195
0.0138 18.6111 670 0.5648 20.2753
0.0133 18.8889 680 0.5655 20.4318
0.0114 19.1667 690 0.5666 20.5569
0.0112 19.4444 700 0.5721 20.3379
0.0108 19.7222 710 0.5714 20.8385
0.0106 20.0 720 0.5744 20.4944
0.0092 20.2778 730 0.5751 20.4318
0.0096 20.5556 740 0.5756 20.3692
0.009 20.8333 750 0.5779 20.1502
0.0084 21.1111 760 0.5790 20.4944
0.0077 21.3889 770 0.5820 20.4005
0.0083 21.6667 780 0.5822 20.4005
0.008 21.9444 790 0.5820 20.4005
0.0077 22.2222 800 0.5829 20.4318
0.0083 22.5 810 0.5843 20.4005
0.0073 22.7778 820 0.5856 20.4005
0.0069 23.0556 830 0.5869 20.4005
0.0067 23.3333 840 0.5886 20.5257
0.007 23.6111 850 0.5882 20.4944
0.0074 23.8889 860 0.5872 20.4631
0.0073 24.1667 870 0.5885 20.4631
0.0066 24.4444 880 0.5896 20.6195
0.0061 24.7222 890 0.5898 20.6195
0.0073 25.0 900 0.5902 20.5882
0.0067 25.2778 910 0.5901 20.6508
0.006 25.5556 920 0.5905 20.5257
0.0061 25.8333 930 0.5911 20.7447
0.0064 26.1111 940 0.5916 20.6821
0.0066 26.3889 950 0.5919 20.6195
0.0071 26.6667 960 0.5924 20.5569
0.006 26.9444 970 0.5923 20.5569
0.0068 27.2222 980 0.5923 20.5257
0.0061 27.5 990 0.5924 20.5257
0.0058 27.7778 1000 0.5924 20.5257

Framework versions

  • Transformers 4.40.1
  • Pytorch 2.2.1+cu121
  • Datasets 2.19.1.dev0
  • Tokenizers 0.19.1