text_shortening_model_v6

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

• Loss: 1.5555
• Rouge1: 0.5993
• Rouge2: 0.3696
• Rougel: 0.551
• Rougelsum: 0.5503
• Bert precision: 0.8968
• Bert recall: 0.9029
• Average word count: 11.2357
• Max word count: 17
• Min word count: 7
• Average token count: 16.4143

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: 0.0003
• train_batch_size: 128
• eval_batch_size: 128
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Bert precision	Bert recall	Average word count	Max word count	Min word count	Average token count
1.2879	1.0	4	1.7189	0.5385	0.3175	0.4882	0.4875	0.8762	0.886	11.8071	18	5	17.1429
1.1303	2.0	8	1.6107	0.5599	0.337	0.5115	0.5117	0.8853	0.8916	11.2071	18	4	16.3071
1.0984	3.0	12	1.5545	0.5828	0.354	0.5254	0.5252	0.8885	0.8985	11.5286	17	4	16.5714
1.052	4.0	16	1.4943	0.5841	0.3631	0.5384	0.5372	0.8917	0.9004	11.3857	17	5	16.6143
0.9922	5.0	20	1.4517	0.5869	0.3671	0.5437	0.5432	0.8912	0.9011	11.5429	17	5	16.7929
0.9524	6.0	24	1.4308	0.5807	0.3571	0.5332	0.5333	0.8883	0.8994	11.6857	17	5	17.0357
0.9008	7.0	28	1.4152	0.5859	0.3585	0.5333	0.5319	0.8885	0.8974	11.4857	17	5	16.7786
0.8787	8.0	32	1.4089	0.5868	0.3592	0.5366	0.5363	0.8901	0.8991	11.4071	17	5	16.8071
0.857	9.0	36	1.4031	0.5974	0.3747	0.5496	0.5494	0.892	0.9015	11.5214	17	5	16.95
0.8122	10.0	40	1.3961	0.5965	0.3716	0.5487	0.5484	0.8917	0.9031	11.7071	17	6	17.1214
0.7943	11.0	44	1.3922	0.6068	0.3774	0.5572	0.5566	0.8947	0.9058	11.5929	17	6	16.9857
0.7632	12.0	48	1.3949	0.6011	0.371	0.55	0.549	0.8944	0.9039	11.4214	16	5	16.9
0.7464	13.0	52	1.3949	0.6007	0.3757	0.5506	0.5492	0.8938	0.9046	11.4357	16	5	16.8714
0.7235	14.0	56	1.3957	0.6113	0.3814	0.5609	0.5601	0.8965	0.9078	11.5429	16	6	16.8714
0.7293	15.0	60	1.3988	0.6102	0.3809	0.5615	0.56	0.8948	0.9079	11.7	16	6	17.15
0.7188	16.0	64	1.3954	0.6094	0.381	0.5603	0.5588	0.8965	0.9062	11.35	16	6	16.8071
0.7028	17.0	68	1.3969	0.6068	0.3846	0.5581	0.5568	0.896	0.9052	11.2571	16	6	16.65
0.6792	18.0	72	1.4056	0.6007	0.3777	0.5519	0.5508	0.895	0.9048	11.3214	16	6	16.6214
0.671	19.0	76	1.4142	0.6043	0.3779	0.5549	0.5541	0.8954	0.9046	11.2429	15	6	16.5429
0.6644	20.0	80	1.4202	0.6009	0.3767	0.5502	0.5496	0.8955	0.9028	11.1643	16	6	16.3643
0.6526	21.0	84	1.4256	0.6023	0.374	0.5485	0.5485	0.8958	0.9032	11.1857	17	6	16.35
0.6311	22.0	88	1.4356	0.6059	0.3768	0.5492	0.5488	0.8932	0.9042	11.5	17	6	16.7214
0.6448	23.0	92	1.4432	0.6071	0.3768	0.5519	0.5518	0.8935	0.9044	11.5357	17	6	16.7643
0.6344	24.0	96	1.4457	0.6088	0.3823	0.5583	0.5576	0.8985	0.9052	11.1214	16	6	16.3071
0.6299	25.0	100	1.4522	0.6049	0.3709	0.5488	0.5484	0.8976	0.9017	10.9	16	6	15.9643
0.6193	26.0	104	1.4616	0.6045	0.3701	0.5499	0.5495	0.8959	0.9032	11.1714	16	6	16.35
0.6247	27.0	108	1.4704	0.5993	0.3719	0.5515	0.5503	0.8949	0.9041	11.3429	17	7	16.6286
0.6062	28.0	112	1.4760	0.6017	0.3702	0.5537	0.5526	0.8949	0.903	11.2929	17	6	16.5143
0.5921	29.0	116	1.4816	0.5994	0.3734	0.5528	0.552	0.8959	0.9025	11.1429	17	6	16.3429
0.5859	30.0	120	1.4887	0.6027	0.3724	0.5523	0.5518	0.8956	0.9034	11.3357	17	7	16.5143
0.5911	31.0	124	1.4958	0.6065	0.3757	0.5523	0.5519	0.8971	0.9033	11.1857	17	6	16.3643
0.5936	32.0	128	1.5029	0.6008	0.3745	0.5508	0.5508	0.8973	0.9015	10.9714	16	6	16.1
0.584	33.0	132	1.5101	0.6087	0.3801	0.5582	0.5583	0.8969	0.9038	11.2214	16	6	16.4071
0.5741	34.0	136	1.5157	0.6054	0.3814	0.5575	0.5576	0.8961	0.9042	11.2643	16	7	16.4786
0.5793	35.0	140	1.5202	0.6079	0.3866	0.5621	0.5622	0.8968	0.9057	11.3214	16	7	16.5714
0.5803	36.0	144	1.5221	0.6081	0.3824	0.5601	0.5602	0.8966	0.9053	11.3357	16	7	16.6214
0.5719	37.0	148	1.5235	0.6025	0.3802	0.555	0.5542	0.898	0.9035	11.1357	16	7	16.3214
0.5567	38.0	152	1.5238	0.5987	0.3763	0.5524	0.5517	0.8974	0.9024	11.0357	16	7	16.2143
0.5535	39.0	156	1.5264	0.6023	0.3746	0.5547	0.5539	0.8977	0.9035	11.1357	16	7	16.3
0.5507	40.0	160	1.5315	0.6039	0.3757	0.5565	0.5559	0.8979	0.9045	11.2071	16	7	16.4143
0.5568	41.0	164	1.5389	0.6078	0.3819	0.5589	0.5579	0.8973	0.9045	11.4	17	7	16.5571
0.5659	42.0	168	1.5444	0.6037	0.3788	0.5567	0.5558	0.8959	0.9036	11.4286	17	7	16.5714
0.561	43.0	172	1.5475	0.5965	0.372	0.5494	0.548	0.8958	0.9024	11.3357	17	7	16.4929
0.5535	44.0	176	1.5493	0.597	0.3703	0.5495	0.5485	0.8967	0.9025	11.2214	17	7	16.3786
0.5542	45.0	180	1.5507	0.6001	0.3706	0.5529	0.5526	0.897	0.9034	11.2429	17	7	16.4214
0.542	46.0	184	1.5527	0.6001	0.3706	0.5529	0.5526	0.897	0.9034	11.2429	17	7	16.4214
0.5466	47.0	188	1.5539	0.6003	0.3702	0.5529	0.5526	0.8968	0.9033	11.2571	17	7	16.4357
0.5478	48.0	192	1.5550	0.5997	0.3699	0.5515	0.5508	0.8969	0.9029	11.2143	17	7	16.3857
0.5429	49.0	196	1.5552	0.5993	0.3696	0.551	0.5503	0.8968	0.9029	11.2357	17	7	16.4143
0.5443	50.0	200	1.5555	0.5993	0.3696	0.551	0.5503	0.8968	0.9029	11.2357	17	7	16.4143

Framework versions

• Transformers 4.32.1
• Pytorch 2.0.1+cu118
• Datasets 2.14.4
• Tokenizers 0.13.3