deed-summarization_version_10

This model is a fine-tuned version of Hasanur525/deed-summarization_version_5 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4150
• Rouge1: 0.3247
• Rouge2: 0.1432
• Rougel: 0.3268
• Rougelsum: 0.3201
• Gen Len: 98.4206

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: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 5000
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
2.6983	1.0	529	1.8154	0.0	0.0	0.0	0.0	72.9565
2.3213	2.0	1058	1.5403	0.0	0.0	0.0	0.0	82.2401
1.0315	3.0	1587	1.2686	0.0	0.0	0.0	0.0	88.1635
1.7308	4.0	2116	1.0681	0.0	0.0	0.0	0.0	89.7155
1.1384	5.0	2645	0.9338	0.0	0.0	0.0	0.0	93.0586
1.6608	6.0	3174	0.8329	0.0199	0.0	0.0199	0.0199	95.5454
1.8287	7.0	3703	0.7506	0.0099	0.0	0.0099	0.0099	96.9036
0.4304	8.0	4232	0.6827	0.0742	0.036	0.0692	0.069	96.8894
1.1026	9.0	4761	0.6189	0.0888	0.0516	0.0888	0.0859	97.5312
0.8345	10.0	5290	0.5662	0.0497	0.0189	0.0443	0.0443	96.8025
0.3368	11.0	5819	0.5291	0.0394	0.0258	0.0398	0.0394	97.9783
0.2668	12.0	6348	0.5010	0.1466	0.0379	0.1368	0.1345	97.4386
0.8294	13.0	6877	0.4787	0.1815	0.0683	0.1744	0.167	97.8639
0.4896	14.0	7406	0.4603	0.1946	0.0732	0.1948	0.1899	97.707
0.4353	15.0	7935	0.4446	0.158	0.0664	0.1476	0.1456	97.8837
1.8165	16.0	8464	0.4314	0.3104	0.1119	0.3005	0.2917	98.4329
0.3503	17.0	8993	0.4236	0.2872	0.1234	0.2785	0.2681	98.2259
0.5756	18.0	9522	0.4199	0.339	0.1242	0.3348	0.3252	98.31
0.7974	19.0	10051	0.4176	0.3437	0.1568	0.3477	0.338	98.3932
0.224	20.0	10580	0.4150	0.3247	0.1432	0.3268	0.3201	98.4206

Framework versions

• Transformers 4.37.2
• Pytorch 2.1.0.dev20230811+cu121
• Datasets 2.17.0
• Tokenizers 0.15.2