itskavya/whisper-large-informal-arabic-new

Whisper Large Informal Arabic

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

• Loss: 0.4559
• Wer: 24.6971
• Cer: 8.2905

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 8
• 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: 100
• training_steps: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.1627	5.2685	200	0.3593	26.4724	9.0815
0.0079	10.5369	400	0.4046	24.8291	8.8585
0.0011	15.8054	600	0.4338	25.0210	8.8444
0.0005	21.0537	800	0.4509	24.6971	8.3210
0.0004	26.3221	1000	0.4559	24.6971	8.2905

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0+cu124
• Datasets 3.4.1
• Tokenizers 0.21.1