arabert_baseline_relevance_task6_fold0

This model is a fine-tuned version of aubmindlab/bert-base-arabertv02 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4526
• Qwk: 0.3226
• Mse: 0.4526

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Qwk	Mse
No log	0.5	2	0.6801	0.0541	0.6801
No log	1.0	4	0.6246	0.1127	0.6246
No log	1.5	6	0.5719	0.2222	0.5719
No log	2.0	8	0.4467	0.2222	0.4467
No log	2.5	10	0.4958	0.3288	0.4958
No log	3.0	12	0.4784	0.4324	0.4784
No log	3.5	14	0.4142	0.5116	0.4142
No log	4.0	16	0.4299	0.5977	0.4299
No log	4.5	18	0.4577	0.58	0.4577
No log	5.0	20	0.4773	0.5664	0.4773
No log	5.5	22	0.4936	0.5664	0.4936
No log	6.0	24	0.4974	0.5000	0.4974
No log	6.5	26	0.5038	0.5088	0.5038
No log	7.0	28	0.5060	0.3937	0.5060
No log	7.5	30	0.4888	0.4444	0.4888
No log	8.0	32	0.4685	0.4444	0.4685
No log	8.5	34	0.4544	0.3226	0.4544
No log	9.0	36	0.4497	0.3226	0.4497
No log	9.5	38	0.4507	0.3226	0.4507
No log	10.0	40	0.4526	0.3226	0.4526

Framework versions

• Transformers 4.44.0
• Pytorch 2.4.0
• Datasets 2.21.0
• Tokenizers 0.19.1