res_nw_gulf_aragpt2-large

This model is a fine-tuned version of aubmindlab/aragpt2-large on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0472
• Bleu: 0.0632
• Rouge1: 0.4039
• Rouge2: 0.1633
• Rougel: 0.4013

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: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 20.0

Training results

Training Loss	Epoch	Step	Bleu	Validation Loss	Rouge1	Rouge2	Rougel
0.2041	1.0	1672	0.0445	0.0487	0.3745	0.1302	0.3718
0.0404	2.0	3344	0.0632	0.0472	0.4039	0.1633	0.4013
0.0301	3.0	5016	0.0763	0.0480	0.4339	0.2002	0.4322
0.0232	4.0	6688	0.0843	0.0515	0.4535	0.2192	0.4517
0.0189	5.0	8360	0.0876	0.0538	0.4654	0.2299	0.4638
0.0164	6.0	10032	0.0930	0.0572	0.4675	0.2370	0.4653
0.0148	7.0	11704	0.0918	0.0583	0.4656	0.2308	0.4636
0.0137	8.0	13376	0.0598	0.0979	0.4720	0.2421	0.4699
0.0128	9.0	15048	0.0623	0.1035	0.4814	0.2488	0.4793
0.0122	10.0	16720	0.0658	0.1046	0.4792	0.2461	0.4778
0.0117	11.0	18392	0.0651	0.1067	0.4881	0.2539	0.4861
0.0112	12.0	20064	0.0677	0.1008	0.4840	0.2490	0.4822

Framework versions

• Transformers 4.45.0.dev0
• Pytorch 2.3.1+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1