Baselhany/Whisper_tiny_with_all_sample_few_epochs

Whisper tiny AR - BH

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

• Loss: 0.0027
• Wer: 0.0379
• Cer: 0.0158

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• 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: 1
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.0001	0.0863	250	0.0024	0.0366	0.0157
0.0	0.1726	500	0.0024	0.0423	0.0214
0.0	0.2589	750	0.0024	0.0357	0.0148
0.0	0.3452	1000	0.0024	0.0423	0.0194
0.0	0.4314	1250	0.0024	0.0638	0.0245
0.0	0.5177	1500	0.0024	0.0634	0.0243
0.0001	0.6040	1750	0.0024	0.0621	0.0236
0.0002	0.6903	2000	0.0024	0.0627	0.0240
0.0002	0.7766	2250	0.0024	0.0622	0.0235
0.0004	0.8629	2500	0.0024	0.0624	0.0237
0.0021	0.9492	2750	0.0024	0.0626	0.0241

Framework versions

• Transformers 4.47.0
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0