arbert_arabic_dialect_identification

This model is a fine-tuned version of lafifi-24/arbert_arabic_dialect_identification on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.7977
• F1-score: 0.5948

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-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	F1-score
No log	1.0	290	1.4459	0.5037
1.4958	2.0	580	1.5787	0.5153
1.4958	3.0	870	1.4938	0.5693
0.7952	4.0	1160	1.7462	0.5504
0.7952	5.0	1450	1.7977	0.5948
0.4922	6.0	1740	2.1508	0.5682
0.3171	7.0	2030	2.5673	0.5335
0.3171	8.0	2320	2.5563	0.5658
0.2204	9.0	2610	2.8487	0.5336
0.2204	10.0	2900	3.0247	0.5422
0.1511	11.0	3190	2.8925	0.5594
0.1511	12.0	3480	3.2729	0.5333
0.0967	13.0	3770	3.2754	0.5531
0.0615	14.0	4060	3.3330	0.5430
0.0615	15.0	4350	3.3549	0.5430
0.043	16.0	4640	3.3918	0.5637
0.043	17.0	4930	3.5727	0.5381
0.037	18.0	5220	3.5450	0.5499
0.0187	19.0	5510	3.5421	0.5394
0.0187	20.0	5800	3.5514	0.5403

Framework versions

• Transformers 4.40.0
• Pytorch 2.2.1+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1