my_awesome_sumarize_model_v2

This model is a fine-tuned version of t5-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8614
• Rouge1: 0.422
• Rouge2: 0.3103
• Rougel: 0.4017
• Rougelsum: 0.4019
• Gen Len: 18.9192

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: 3.419313942464226e-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: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	1.0	239	1.0311	0.418	0.304	0.3985	0.3988	18.9267
No log	2.0	478	1.0058	0.4198	0.3065	0.4001	0.4004	18.9229
1.1809	3.0	717	0.9693	0.4215	0.3085	0.402	0.4024	18.9192
1.1809	4.0	956	0.9489	0.4208	0.3068	0.4016	0.402	18.9211
1.0899	5.0	1195	0.9402	0.4208	0.3074	0.4015	0.4019	18.9211
1.0899	6.0	1434	0.9204	0.4239	0.3125	0.4046	0.4048	18.9135
1.0455	7.0	1673	0.9111	0.4223	0.3094	0.4023	0.4024	18.9173
1.0455	8.0	1912	0.9055	0.4219	0.3106	0.4022	0.4024	18.9173
1.01	9.0	2151	0.8958	0.4218	0.3106	0.4016	0.4019	18.9154
1.01	10.0	2390	0.8901	0.4213	0.3106	0.4017	0.4022	18.9173
0.9841	11.0	2629	0.8828	0.4221	0.3117	0.4024	0.4029	18.9154
0.9841	12.0	2868	0.8749	0.4217	0.3102	0.4018	0.4021	18.9173
0.9599	13.0	3107	0.8755	0.4217	0.3104	0.4019	0.4023	18.9173
0.9599	14.0	3346	0.8733	0.4214	0.3103	0.4015	0.4016	18.9173
0.9487	15.0	3585	0.8701	0.4215	0.3097	0.4017	0.4019	18.9192
0.9487	16.0	3824	0.8663	0.4213	0.3099	0.4013	0.4016	18.9192
0.9396	17.0	4063	0.8647	0.4215	0.3092	0.4013	0.4015	18.9192
0.9396	18.0	4302	0.8621	0.4218	0.3098	0.4015	0.4018	18.9192
0.9329	19.0	4541	0.8615	0.422	0.3103	0.4017	0.4019	18.9192
0.9329	20.0	4780	0.8614	0.422	0.3103	0.4017	0.4019	18.9192

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2