raghadalghonaim/whisper-small-ar

whisper_small_ar_ralghonaim

This model is a fine-tuned version of openai/whisper-small on the My Own Arabic DS dataset. It achieves the following results on the evaluation set:

• Loss: 0.5957
• Wer: 42.6937

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: 16
• eval_batch_size: 8
• 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: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.6446	0.9268	1000	0.6416	48.7542
0.4115	1.8536	2000	0.5765	44.4449
0.2475	2.7804	3000	0.5745	43.3860
0.1478	3.7071	4000	0.5957	42.6937

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1