_Arabic_nougat_AHR

This model is a fine-tuned version of MohamedRashad/arabic-base-nougat on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3533

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: 0.0001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 6
• total_train_batch_size: 48
• 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
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss
2.3594	1.0	77	0.3758
1.6401	2.0	154	0.3143
1.4332	3.0	231	0.2986
0.9917	4.0	308	0.2993
0.8202	5.0	385	0.3082
0.669	6.0	462	0.3103
0.5788	7.0	539	0.3233
0.4873	8.0	616	0.3336
0.4865	9.0	693	0.3366
0.3386	10.0	770	0.3503
0.3643	11.0	847	0.3476
0.3229	12.0	924	0.3546
0.3406	13.0	1001	0.3536
0.331	14.0	1078	0.3534
0.3016	14.8140	1140	0.3533

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0