results_arat5-3_wiki

This model is a fine-tuned version of hiba2/results_arat5-2_wiki on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.7821
• Rouge1: 0.0926
• Rouge2: 0.0015
• Rougel: 0.0934
• Rougelsum: 0.0928
• Gen Len: 19.0

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
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Gen Len	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum
7.8936	0.9506	500	0.0	5.9107	0.0	0.0	0.0	0.0
5.5649	1.9011	1000	18.8876	4.9336	0.0905	0.0	0.0915	0.0912
4.9098	2.8517	1500	18.8989	4.3731	0.0905	0.0	0.0915	0.0912
4.4486	3.8023	2000	19.0	3.9340	0.0875	0.0	0.0885	0.0882
4.0755	4.7529	2500	18.9382	3.5412	0.0881	0.0	0.0891	0.0887
3.6998	5.7034	3000	18.8783	3.1344	0.095	0.0002	0.0958	0.0954
3.3129	6.6540	3500	18.8408	2.8528	0.0935	0.0013	0.0945	0.094
3.1053	7.6046	4000	18.9382	2.6196	0.0936	0.0008	0.0946	0.0941
2.8412	8.5551	4500	18.867	2.4414	0.091	0.0011	0.0919	0.0915
2.702	9.5057	5000	18.8783	2.2952	0.0936	0.001	0.0948	0.0946
2.5611	10.4563	5500	19.0	2.1816	0.093	0.0011	0.0941	0.0936
2.4499	11.4068	6000	18.8502	2.0914	0.0988	0.0011	0.0995	0.099
2.3764	12.3574	6500	18.8371	2.0264	0.0992	0.0016	0.0997	0.0995
2.3172	13.3080	7000	18.9888	1.9853	0.098	0.0015	0.099	0.0986
2.2794	14.2586	7500	18.9888	1.9615	0.0971	0.0023	0.0977	0.0976
2.2178	15.2091	8000	1.9424	0.0961	0.0009	0.0972	0.0968	19.0
2.2378	16.1597	8500	1.8855	0.0935	0.0011	0.0942	0.0937	19.0
2.1573	17.1103	9000	1.8386	0.0952	0.0009	0.0962	0.0958	19.0
2.132	18.0608	9500	1.8055	0.0919	0.0012	0.0929	0.0923	18.8783
2.1035	19.0114	10000	1.7863	0.0942	0.0015	0.0949	0.0945	19.0
2.0818	19.9620	10500	1.7821	0.0926	0.0015	0.0934	0.0928	19.0

Framework versions

• Transformers 4.42.0.dev0
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1