whisper-small-ar-eos-v8-eos-v12-2

This model is a fine-tuned version of Ahmed107/whisper-small-ar-eos-v8 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 3.1011
• Accuracy: 0.6791

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: 32
• eval_batch_size: 32
• seed: 42
• 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_ratio: 0.2
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.6108	1.0	506	0.7022	0.5
0.5868	2.0	1012	0.6366	0.6780
0.4895	3.0	1518	0.6820	0.6523
0.5727	4.0	2024	0.5976	0.6942
0.4475	5.0	2530	0.6062	0.6992
0.3666	6.0	3036	0.6934	0.6858
0.3592	7.0	3542	0.7931	0.6864
0.1915	8.0	4048	0.9717	0.6747
0.0762	9.0	4554	1.2450	0.6797
0.0433	10.0	5060	1.6070	0.6886
0.1796	11.0	5566	2.1124	0.6786
0.0163	12.0	6072	2.1746	0.6741
0.012	13.0	6578	2.5129	0.6775
0.0639	14.0	7084	2.6168	0.6858
0.0378	15.0	7590	2.5820	0.6869
0.0001	16.0	8096	2.7996	0.6741
0.0081	17.0	8602	2.8051	0.6836
0.0001	18.0	9108	2.8716	0.6875
0.0	19.0	9614	3.0780	0.6702
0.0597	20.0	10120	2.7799	0.6925
0.0538	21.0	10626	2.7572	0.6847
0.0	22.0	11132	2.7749	0.6897
0.0	23.0	11638	2.8876	0.6881
0.0	24.0	12144	2.9893	0.6724
0.0	25.0	12650	3.0065	0.6802
0.0	26.0	13156	3.0087	0.6825
0.0	27.0	13662	3.0478	0.6786
0.0	28.0	14168	3.0592	0.6780
0.0	29.0	14674	3.0902	0.6775
0.0	30.0	15180	3.1011	0.6791

Framework versions

• Transformers 4.48.0.dev0
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0