Whisper base AR - YA

This model is a fine-tuned version of openai/whisper-base on the quran-ayat-speech-to-text dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0033
  • Wer: 0.0497
  • Cer: 0.0200

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: 0.0001
  • train_batch_size: 8
  • eval_batch_size: 8
  • 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
  • lr_scheduler_warmup_steps: 500
  • num_epochs: 25
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer Cer
0.0024 1.0 320 0.0034 0.0440 0.0180
0.0014 2.0 640 0.0049 0.0653 0.0257
0.0013 3.0 960 0.0057 0.0766 0.0283
0.0007 4.0 1280 0.0057 0.0681 0.0290
0.0004 5.0 1600 0.0057 0.0617 0.0253
0.0002 6.0 1920 0.0060 0.0662 0.0244
0.0002 7.0 2240 0.0068 0.0624 0.0237
0.0003 8.0 2560 0.0061 0.0652 0.0259
0.0003 9.0 2880 0.0067 0.0648 0.0252
0.0004 10.0 3200 0.0062 0.0670 0.0259
0.0002 11.0 3520 0.0061 0.0610 0.0230
0.0001 12.0 3840 0.0064 0.0581 0.0217
0.0001 13.0 4160 0.0061 0.0576 0.0217
0.0 14.0 4480 0.0062 0.0594 0.0235
0.0 15.0 4800 0.0066 0.0630 0.0251
0.0 16.0 5120 0.0069 0.0581 0.0240
0.0 17.0 5440 0.0070 0.0579 0.0228
0.0 18.0 5760 0.0071 0.0586 0.0232
0.0 19.0 6080 0.0072 0.0590 0.0239
0.0 20.0 6400 0.0072 0.0576 0.0234
0.0 21.0 6720 0.0073 0.0574 0.0239
0.0 22.0 7040 0.0073 0.0577 0.0240
0.0 23.0 7360 0.0074 0.0577 0.0240
0.0 24.0 7680 0.0076 0.0613 0.0246
0.0 25.0 8000 0.0074 0.0581 0.0244

Framework versions

  • Transformers 4.46.3
  • Pytorch 2.4.0
  • Datasets 3.1.0
  • Tokenizers 0.20.3
Downloads last month
1,398
Safetensors
Model size
72.6M params
Tensor type
F32
·
Inference Providers NEW
This model is not currently available via any of the supported Inference Providers.

Model tree for YoussefAshmawy/Graduation_Project_Whisper_base

Finetuned
(435)
this model