quran-asr

This model is a fine-tuned version of uzair0/quran-asr on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.0083
• Wer: 0.6094
• Cer: 0.1724

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: 1
• eval_batch_size: 1
• seed: 1869940075
• gradient_accumulation_steps: 16
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.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: 800
• training_steps: 15000

Training results

Training Loss	Epoch	Step	Cer	Validation Loss	Wer
0.3657	4.5669	2000	0.2276	1.0013	0.759
0.2776	9.1326	4000	0.2153	0.9888	0.7128
0.2465	13.6994	6000	0.201	1.0453	0.7355
0.1763	18.2651	8000	0.1908	1.0804	0.733
0.1584	22.832	10000	0.1915	1.0795	0.7153
0.1762	27.3977	12000	1.0009	0.6368	0.1796
0.1373	31.9646	14000	1.0083	0.6094	0.1724

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 4.8.3
• Tokenizers 0.22.2