AraT5-msa-small-finetuned-xlsum-ar

This model is a fine-tuned version of UBC-NLP/AraT5-msa-small on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 5.0192
• Rouge1: 10.2878
• Rouge2: 2.314
• Rougel: 9.2899
• Rougelsum: 9.308
• Gen Len: 18.9057

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
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
7.9587	1.0	2111	7.2900	5.9427	0.2445	5.6768	5.6871	18.7825
7.1801	2.0	4222	6.4007	5.2133	0.4475	4.8073	4.8155	18.557
6.6971	3.0	6333	5.9127	5.974	1.0053	5.5186	5.5093	18.9638
6.396	4.0	8444	5.5987	7.5554	1.4469	6.8972	6.9011	18.912
6.1389	5.0	10555	5.3854	8.3705	1.811	7.6315	7.6573	18.9488
6.0006	6.0	12666	5.2364	9.5287	2.0941	8.6128	8.6469	18.9499
5.9139	7.0	14777	5.1351	10.073	2.3252	9.0992	9.1164	18.9254
5.8562	8.0	16888	5.0689	10.3972	2.4912	9.3707	9.3923	18.9211
5.7958	9.0	18999	5.0295	10.6349	2.5819	9.5917	9.6156	18.9179
5.7783	10.0	21110	5.0182	10.6062	2.5793	9.5691	9.5961	18.9163

Framework versions

• Transformers 4.39.3
• Pytorch 2.1.2
• Datasets 2.18.0
• Tokenizers 0.15.2