t5-base-mse-summarization

This model is a fine-tuned version of t5-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.8743
• Rouge1: 45.9597
• Rouge2: 26.8086
• Rougel: 39.935
• Rougelsum: 43.8897
• Bleurt: -0.7132
• Gen Len: 18.464

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: 1e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• 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	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Bleurt	Gen Len
1.2568	1.0	267	1.0472	41.6829	21.9654	35.4264	39.5556	-0.8231	18.522
1.1085	2.0	534	0.9840	43.1479	23.3351	36.9244	40.886	-0.7843	18.534
1.0548	3.0	801	0.9515	44.1511	24.4912	37.9549	41.9984	-0.7702	18.528
1.0251	4.0	1068	0.9331	44.426	24.9439	38.2978	42.1731	-0.7633	18.619
0.9888	5.0	1335	0.9201	45.0385	25.524	38.8681	42.8998	-0.7497	18.523
0.9623	6.0	1602	0.9119	44.8648	25.469	38.9281	42.7798	-0.7496	18.537
0.9502	7.0	1869	0.9015	44.9668	25.5041	38.9463	42.9368	-0.7412	18.48
0.9316	8.0	2136	0.8973	45.3028	25.7232	39.1533	43.277	-0.7318	18.523
0.9191	9.0	2403	0.8921	45.2901	25.916	39.2909	43.3022	-0.7296	18.529
0.9122	10.0	2670	0.8889	45.3535	26.1369	39.4861	43.28	-0.7271	18.545
0.8993	11.0	2937	0.8857	45.5345	26.1669	39.5656	43.4664	-0.7269	18.474
0.8905	12.0	3204	0.8816	45.7796	26.4145	39.8117	43.734	-0.7185	18.503
0.8821	13.0	3471	0.8794	45.7163	26.4314	39.719	43.6407	-0.7211	18.496
0.8789	14.0	3738	0.8784	45.9097	26.7281	39.9071	43.8105	-0.7127	18.452
0.8665	15.0	4005	0.8765	46.1148	26.8882	40.1006	43.988	-0.711	18.443
0.8676	16.0	4272	0.8766	45.9119	26.7674	39.9001	43.8237	-0.718	18.491
0.8637	17.0	4539	0.8758	45.9158	26.7153	39.9463	43.8323	-0.7183	18.492
0.8622	18.0	4806	0.8752	45.9508	26.75	39.9533	43.8795	-0.7144	18.465
0.8588	19.0	5073	0.8744	45.9192	26.7352	39.8921	43.8204	-0.7148	18.462
0.8554	20.0	5340	0.8743	45.9597	26.8086	39.935	43.8897	-0.7132	18.464

Framework versions

• Transformers 4.21.2
• Pytorch 1.12.1+cu113
• Datasets 2.4.0
• Tokenizers 0.12.1