text_shortening_model_v79

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

• Loss: 1.0551
• Bert precision: 0.8947
• Bert recall: 0.8962
• Bert f1-score: 0.895
• Average word count: 6.7804
• Max word count: 16
• Min word count: 1
• Average token count: 10.8466
• % shortened texts with length > 12: 1.5951

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: 7e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Bert precision	Bert recall	Bert f1-score	Average word count	Max word count	Min word count	Average token count	% shortened texts with length > 12
2.0194	1.0	30	1.4487	0.8778	0.8746	0.8755	6.7755	16	1	10.7288	2.3313
1.58	2.0	60	1.3193	0.8835	0.8837	0.883	6.9301	16	2	10.7791	2.3313
1.4385	3.0	90	1.2492	0.8833	0.8855	0.8839	7.0368	16	2	10.9816	2.6994
1.3616	4.0	120	1.2111	0.8877	0.8873	0.887	6.8466	16	2	10.7509	1.8405
1.2976	5.0	150	1.1685	0.8869	0.8878	0.8868	6.8564	17	2	10.8172	1.8405
1.2495	6.0	180	1.1559	0.8885	0.8895	0.8885	6.8577	16	2	10.8564	2.0859
1.201	7.0	210	1.1353	0.8889	0.891	0.8894	6.9521	16	2	11.0012	2.3313
1.1717	8.0	240	1.1164	0.8892	0.89	0.8891	6.8601	16	1	10.8933	2.0859
1.1352	9.0	270	1.1110	0.8902	0.8891	0.8891	6.708	16	1	10.7436	1.1043
1.0984	10.0	300	1.1037	0.8901	0.8909	0.8901	6.8233	17	1	10.8503	1.9632
1.0745	11.0	330	1.0937	0.8894	0.892	0.8902	6.9362	17	2	10.9742	2.3313
1.0509	12.0	360	1.0907	0.8911	0.8916	0.8908	6.8233	17	1	10.8564	1.9632
1.0269	13.0	390	1.0805	0.8906	0.8934	0.8915	6.9448	17	1	11.0135	2.2086
1.0126	14.0	420	1.0784	0.8912	0.8935	0.8919	6.9264	17	2	10.973	2.3313
0.9959	15.0	450	1.0725	0.8929	0.8944	0.8932	6.8294	17	1	10.8957	2.2086
0.9717	16.0	480	1.0715	0.8916	0.8941	0.8924	6.919	17	1	10.9963	2.0859
0.9552	17.0	510	1.0727	0.8935	0.8949	0.8937	6.8282	17	1	10.9055	1.9632
0.9461	18.0	540	1.0665	0.8947	0.8955	0.8947	6.8061	17	1	10.8613	1.5951
0.926	19.0	570	1.0664	0.8948	0.896	0.895	6.7853	16	1	10.8515	1.3497
0.9192	20.0	600	1.0636	0.8948	0.8953	0.8946	6.7718	16	1	10.8209	1.4724
0.9101	21.0	630	1.0581	0.8954	0.897	0.8957	6.8221	16	1	10.8724	1.5951
0.899	22.0	660	1.0599	0.8954	0.8974	0.8959	6.8405	16	1	10.8982	1.5951
0.8843	23.0	690	1.0586	0.8943	0.8962	0.8948	6.8393	17	2	10.9055	1.9632
0.8779	24.0	720	1.0572	0.8932	0.8961	0.8942	6.8736	17	2	10.9656	2.0859
0.8725	25.0	750	1.0573	0.8939	0.8963	0.8947	6.8098	16	2	10.9104	1.7178
0.8567	26.0	780	1.0591	0.8951	0.8968	0.8955	6.7926	17	1	10.8945	1.5951
0.8549	27.0	810	1.0577	0.8945	0.8962	0.8948	6.8135	17	1	10.9018	1.8405
0.8467	28.0	840	1.0570	0.8948	0.8961	0.895	6.7669	16	1	10.8405	1.4724
0.833	29.0	870	1.0577	0.895	0.896	0.895	6.7546	16	1	10.8294	1.3497
0.8284	30.0	900	1.0548	0.8942	0.8957	0.8945	6.7816	16	1	10.8589	1.4724
0.8296	31.0	930	1.0565	0.8947	0.8967	0.8952	6.8037	16	1	10.8982	1.4724
0.8156	32.0	960	1.0550	0.8945	0.8961	0.8948	6.7914	16	2	10.8601	1.5951
0.8095	33.0	990	1.0567	0.8944	0.8962	0.8948	6.8049	16	2	10.881	1.7178
0.8066	34.0	1020	1.0564	0.8948	0.8961	0.895	6.7853	16	1	10.8405	1.8405
0.817	35.0	1050	1.0567	0.8951	0.8961	0.8952	6.7509	16	1	10.8172	1.5951
0.8155	36.0	1080	1.0563	0.8949	0.8964	0.8952	6.7669	16	1	10.838	1.5951
0.808	37.0	1110	1.0560	0.8946	0.8965	0.8951	6.7926	16	1	10.8675	1.7178
0.8049	38.0	1140	1.0554	0.895	0.8965	0.8953	6.7742	16	1	10.8393	1.4724
0.8002	39.0	1170	1.0550	0.8946	0.8962	0.8949	6.7877	16	1	10.8491	1.5951
0.7912	40.0	1200	1.0551	0.8947	0.8962	0.895	6.7804	16	1	10.8466	1.5951

Framework versions

• Transformers 4.33.1
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.13.3