Hugging Face's logo

makhataei
/
Whisper-Small-Common-Voice

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
Persian
whisper
fa-asr
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
Whisper-Small-Common-Voice / README.md
makhataei's picture
makhataei
End of training
814c436
preview code
|
raw
history blame
7.78 kB
metadata
language:
  - fa
license: apache-2.0
base_model: makhataei/Whisper-Small-Common-Voice
tags:
  - fa-asr
  - generated_from_trainer
datasets:
  - mozilla-foundation/common_voice_15_0
metrics:
  - wer
model-index:
  - name: Whisper Small Persian
    results: []

Whisper Small Persian

This model is a fine-tuned version of makhataei/Whisper-Small-Common-Voice on the Common Voice 15.0 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.8843
  • Wer: 48.6448

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: 14
  • eval_batch_size: 10
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 56
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 50
  • training_steps: 10000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
0.1801 0.39 100 0.4976 49.1260
0.1597 0.79 200 0.4624 46.7497
0.0776 1.18 300 0.4794 43.1761
0.083 1.57 400 0.4823 43.8028
0.0786 1.96 500 0.4883 44.3915
0.0331 2.36 600 0.5385 46.2437
0.0353 2.75 700 0.5605 44.9439
0.0139 3.14 800 0.5941 45.2812
0.0152 3.53 900 0.5978 49.2930
0.0155 3.93 1000 0.6114 49.9777
0.0063 4.32 1100 0.6467 50.0041
0.0079 4.71 1200 0.6383 48.0875
0.0046 5.1 1300 0.6500 45.4995
0.0042 5.5 1400 0.6476 47.4492
0.0052 5.89 1500 0.6685 52.1870
0.0023 6.28 1600 0.6794 44.2510
0.0032 6.67 1700 0.6724 45.7161
0.0021 7.07 1800 0.6820 47.6013
0.0015 7.46 1900 0.6925 46.6720
0.0024 7.85 2000 0.7104 50.2902
0.0029 8.24 2100 0.6837 46.4173
0.0016 8.64 2200 0.7191 46.0088
0.0017 9.03 2300 0.7105 47.5964
0.0014 9.42 2400 0.7293 44.7603
0.0018 9.81 2500 0.7365 49.8966
0.0008 10.21 2600 0.7378 47.4740
0.0016 10.6 2700 0.7303 45.9691
0.0011 10.99 2800 0.7330 47.7254
0.0014 11.38 2900 0.7448 44.8579
0.0013 11.78 3000 0.7471 46.5116
0.0015 12.17 3100 0.7513 47.5699
0.0014 12.56 3200 0.7380 46.4008
0.0015 12.95 3300 0.7520 45.9658
0.0009 13.35 3400 0.7482 49.2269
0.0021 13.74 3500 0.7619 47.1234
0.0013 14.13 3600 0.7453 49.3956
0.001 14.52 3700 0.7582 47.6741
0.0009 14.92 3800 0.7637 46.9713
0.0014 15.31 3900 0.7722 47.2706
0.001 15.7 4000 0.7692 46.9663
0.0003 16.09 4100 0.7744 47.1730
0.0004 16.49 4200 0.7842 47.3351
0.0003 16.88 4300 0.7784 47.0771
0.0002 17.27 4400 0.7879 45.7641
0.0005 17.66 4500 0.7965 50.0240
0.0004 18.06 4600 0.8001 48.4381
0.0002 18.45 4700 0.8016 49.0037
0.0002 18.84 4800 0.8066 50.0868
0.0009 19.23 4900 0.8021 47.2276
0.0005 19.63 5000 0.8162 47.3500
0.0003 20.02 5100 0.8091 48.7225
0.0003 20.41 5200 0.8060 51.5024
0.0003 20.8 5300 0.8220 51.4875
0.0003 21.2 5400 0.8098 45.8617
0.0003 21.59 5500 0.8132 44.8711
0.0009 21.98 5600 0.8006 45.3937
0.0003 22.37 5700 0.8008 45.6186
0.0002 22.77 5800 0.8081 46.3247
0.0002 23.16 5900 0.8082 46.1279
0.0002 23.55 6000 0.8238 46.1775
0.0005 23.95 6100 0.8119 49.9727
0.0002 24.34 6200 0.8315 49.0863
0.0001 24.73 6300 0.8224 47.2243
0.0001 25.12 6400 0.8259 47.1681
0.0001 25.52 6500 0.8219 48.5737
0.0002 25.91 6600 0.8400 48.9077
0.0005 26.3 6700 0.8319 47.5567
0.0001 26.69 6800 0.8394 50.2357
0.0001 27.09 6900 0.8480 48.4629
0.0001 27.48 7000 0.8498 47.1151
0.0002 27.87 7100 0.8342 48.9243
0.0003 28.26 7200 0.8184 47.3731
0.0001 28.66 7300 0.8278 47.9288
0.0002 29.05 7400 0.8439 47.8610
0.0001 29.44 7500 0.8461 49.9463
0.0001 29.83 7600 0.8449 48.4861
0.0001 30.23 7700 0.8512 49.0003
0.0001 30.62 7800 0.8555 48.2777
0.0001 31.01 7900 0.8543 48.6795
0.0001 31.4 8000 0.8566 48.7655
0.0001 31.8 8100 0.8605 48.6779
0.0 32.19 8200 0.8634 49.3691
0.0 32.58 8300 0.8663 50.0438
0.0 32.97 8400 0.8685 49.7280
0.0 33.37 8500 0.8704 49.1641
0.0 33.76 8600 0.8724 48.8416
0.0 34.15 8700 0.8736 49.2286
0.0 34.54 8800 0.8755 48.6134
0.0 34.94 8900 0.8767 48.9259
0.0 35.33 9000 0.8778 48.9805
0.0 35.72 9100 0.8791 49.3212
0.0 36.11 9200 0.8801 49.3724
0.0 36.51 9300 0.8813 49.4336
0.0 36.9 9400 0.8819 49.1045
0.0 37.29 9500 0.8826 49.2633
0.0 37.68 9600 0.8832 49.4237
0.0 38.08 9700 0.8837 48.6316
0.0 38.47 9800 0.8841 48.6465
0.0 38.86 9900 0.8842 48.9342
0.0 39.25 10000 0.8843 48.6448

Framework versions

  • Transformers 4.35.0
  • Pytorch 2.0.1+cu117
  • Datasets 2.14.6
  • Tokenizers 0.14.1