Mohamedshaaban2001/MSDC-whisper-large-v3-56

Whisper whisper-large-v3 ar1 - Mohamed Shaaban

This model is a fine-tuned version of openai/whisper-large-v3 on the Common standard ar Voice 11.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.4220
• Wer: 0.0

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
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.5721	1.0	1	2.1602	100.0
0.5723	2.0	2	1.0610	33.3333
0.1861	3.0	3	0.6003	33.3333
0.0478	4.0	4	0.4661	0.0
0.0262	5.0	5	0.4220	0.0

Framework versions

• Transformers 4.39.3
• Pytorch 2.2.2
• Datasets 2.18.0
• Tokenizers 0.15.2