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
d68af68 verified
preview code
|
raw
history blame
No virus
3.96 kB
metadata
language:
  - fa
license: apache-2.0
base_model: makhataei/Whisper-Small-Common-Voice
tags:
  - fa-asr
  - generated_from_trainer
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 Ctejarat dataset. It achieves the following results on the evaluation set:

  • Loss: 0.5349
  • Wer: 26.2116

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-06
  • train_batch_size: 10
  • eval_batch_size: 10
  • seed: 42
  • gradient_accumulation_steps: 8
  • total_train_batch_size: 80
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 100
  • training_steps: 4000

Training results

Training Loss Epoch Step Validation Loss Wer
0.2944 9.64 100 0.4843 33.6519
0.1048 19.28 200 0.4394 30.1706
0.0273 28.92 300 0.4493 29.7611
0.0083 38.55 400 0.4645 29.4198
0.0042 48.19 500 0.4744 28.5324
0.0026 57.83 600 0.4811 28.3276
0.0018 67.47 700 0.4863 27.6451
0.0014 77.11 800 0.4907 27.7816
0.0012 86.75 900 0.4945 27.4403
0.0009 96.39 1000 0.4979 27.4403
0.0008 106.02 1100 0.5010 26.8259
0.0007 115.66 1200 0.5036 26.8259
0.0006 125.3 1300 0.5062 26.6894
0.0006 134.94 1400 0.5085 26.3481
0.0005 144.58 1500 0.5107 26.3481
0.0004 154.22 1600 0.5126 26.4164
0.0004 163.86 1700 0.5145 26.4846
0.0004 173.49 1800 0.5163 26.3481
0.0003 183.13 1900 0.5179 30.8532
0.0003 192.77 2000 0.5194 30.8532
0.0003 202.41 2100 0.5209 30.7850
0.0003 212.05 2200 0.5222 30.9215
0.0003 221.69 2300 0.5236 30.9215
0.0003 231.33 2400 0.5248 30.9215
0.0002 240.96 2500 0.5259 30.9215
0.0002 250.6 2600 0.5270 30.7167
0.0002 260.24 2700 0.5280 30.8532
0.0002 269.88 2800 0.5290 30.8532
0.0002 279.52 2900 0.5299 30.7167
0.0002 289.16 3000 0.5306 30.7167
0.0002 298.8 3100 0.5314 30.7167
0.0002 308.43 3200 0.5321 30.7167
0.0002 318.07 3300 0.5327 30.7850
0.0002 327.71 3400 0.5333 30.7167
0.0002 337.35 3500 0.5337 30.7167
0.0002 346.99 3600 0.5341 30.7167
0.0002 356.63 3700 0.5344 30.6485
0.0002 366.27 3800 0.5347 26.2116
0.0002 375.9 3900 0.5348 26.2116
0.0002 385.54 4000 0.5349 26.2116

Framework versions

  • Transformers 4.35.2
  • Pytorch 2.0.1+cu117
  • Datasets 2.15.0
  • Tokenizers 0.15.0