Arabizi_Checkpoints_vlast_d5

This model is a fine-tuned version of Mohamedd123321/Arabizi_Checkpoints-large on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.7717
• Accuracy: 0.8692
• F1: 0.8708
• Precision: 0.8747
• Recall: 0.8692

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: 8
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.06
• num_epochs: 6

Training results

Training Loss	Epoch	Step	Accuracy	F1	Validation Loss	Precision	Recall
0.4211	0.6906	2000	0.8541	0.8589	0.3843	0.8695	0.8541
0.3387	1.3812	4000	0.8618	0.8650	0.4193	0.8725	0.8618
0.2823	2.0718	6000	0.8621	0.8657	0.4879	0.8750	0.8621
0.2742	2.7624	8000	0.8609	0.8648	0.4871	0.8742	0.8609
0.2181	3.4530	10000	0.8650	0.8672	0.6036	0.8739	0.8650
0.1666	4.1436	12000	0.8689	0.8717	0.6616	0.8780	0.8689
0.1701	4.8343	14000	0.7497	0.8713	0.8727	0.8752	0.8713
0.1382	5.5249	16000	0.7717	0.8692	0.8708	0.8747	0.8692

Framework versions

• Transformers 4.40.2
• Pytorch 2.8.0+cu129
• Datasets 5.0.0
• Tokenizers 0.19.1