ahishamm/finetuned-whisper-quranic-large-v3-revised

QURANIC Whisper Large V3 - revised

This model is a fine-tuned version of openai/whisper-large-v3 on the Common_voice_16_1 dataset. It achieves the following results on the evaluation set:

• Loss: 0.2252
• Wer: 163.3859

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: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 20000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.3095	0.21	2000	0.3293	155.0801
0.2412	0.41	4000	0.3059	287.9687
0.1762	0.62	6000	0.2843	152.7845
0.1906	0.82	8000	0.2645	124.8897
0.0952	1.03	10000	0.2535	129.0233
0.0955	1.24	12000	0.2567	141.4259
0.0865	1.44	14000	0.2360	205.5690
0.1363	1.65	16000	0.2288	187.0938
0.1038	1.86	18000	0.2197	178.2311
0.062	2.06	20000	0.2252	163.3859

Framework versions

• Transformers 4.39.2
• Pytorch 2.2.0
• Datasets 2.18.0
• Tokenizers 0.15.1