abdullah090809/whisper-medium-ur-fleurs-v2

Whisper Medium Ur - Muhammad Abdullah on Fleurs

This model is a fine-tuned version of GogetaBlueMUI/whisper-medium-ur-jalandhary on the Fleurs Urdu dataset. It achieves the following results on the evaluation set:

• Loss: 0.4401
• Wer: 23.0050

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: 5e-06
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• 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: 60
• training_steps: 600
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.1051	0.7576	100	0.3871	23.5851
0.0437	1.5152	200	0.4073	23.7390
0.0209	2.2727	300	0.4184	23.1944
0.0199	3.0303	400	0.4228	23.0523
0.009	3.7879	500	0.4347	23.3602
0.0048	4.5455	600	0.4401	23.0050

Framework versions

• Transformers 4.49.0
• Pytorch 2.5.1+cu121
• Datasets 3.3.2
• Tokenizers 0.21.0