results

This model is a fine-tuned version of malmarjeh/t5-arabic-text-summarization on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0104
• Rouge1: 0.1382
• Rouge2: 0.0187
• Rougel: 0.1382
• Rougelsum: 0.1382
• Gen Len: 18.9404

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: 0.0005
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
0.0338	0.23	500	0.0175	0.1514	0.0297	0.1511	0.1518	18.9188
0.0566	0.46	1000	0.0161	0.1565	0.0388	0.157	0.1573	18.9188
0.0418	0.7	1500	0.0125	0.1372	0.0199	0.1375	0.1379	18.8105
0.0333	0.93	2000	0.0116	0.1443	0.0253	0.1448	0.1448	18.8051
0.0287	1.16	2500	0.0110	0.144	0.0192	0.1442	0.1442	19.0
0.0247	1.39	3000	0.0096	0.1511	0.024	0.1517	0.1518	19.0
0.0219	1.62	3500	0.0087	0.1463	0.0241	0.1462	0.1462	18.9747
0.021	1.86	4000	0.0104	0.1382	0.0187	0.1382	0.1382	18.9404

Framework versions

• Transformers 4.39.0.dev0
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2