text_shortening_model_v76

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

• Loss: 1.1244
• Bert precision: 0.8967
• Bert recall: 0.8969
• Bert f1-score: 0.8964
• Average word count: 6.8061
• Max word count: 16
• Min word count: 2
• Average token count: 10.9902
• % 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: 0.0001
• 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
1.8741	1.0	30	1.3873	0.8846	0.8811	0.8823	6.7558	15	2	10.6282	2.5767
1.4617	2.0	60	1.2781	0.8879	0.8867	0.8868	6.8613	16	2	10.773	0.9816
1.3352	3.0	90	1.2202	0.8908	0.8894	0.8896	6.8503	14	2	10.8245	0.9816
1.2484	4.0	120	1.1879	0.892	0.8902	0.8907	6.7816	17	1	10.7963	1.1043
1.1842	5.0	150	1.1657	0.893	0.8904	0.8913	6.6945	14	2	10.6822	0.6135
1.1263	6.0	180	1.1490	0.8932	0.8921	0.8921	6.8601	17	2	10.8663	1.7178
1.0859	7.0	210	1.1347	0.8909	0.8942	0.8921	7.0663	17	1	11.1975	2.3313
1.0511	8.0	240	1.1219	0.8925	0.8934	0.8925	6.865	17	1	11.0074	1.227
1.0023	9.0	270	1.1118	0.8936	0.8937	0.8931	6.8393	17	1	10.9963	1.7178
0.9795	10.0	300	1.1073	0.8939	0.8929	0.8929	6.7227	17	1	10.8528	0.8589
0.9489	11.0	330	1.1050	0.8932	0.8951	0.8937	6.9337	17	2	11.0969	1.5951
0.9275	12.0	360	1.1026	0.8945	0.8953	0.8945	6.8331	17	2	11.0135	1.4724
0.8829	13.0	390	1.0989	0.8946	0.8957	0.8947	6.8638	17	1	11.038	1.3497
0.8762	14.0	420	1.0975	0.8939	0.8962	0.8946	6.9239	17	1	11.1423	2.0859
0.8559	15.0	450	1.0988	0.8953	0.8953	0.8948	6.8049	16	1	10.9742	1.7178
0.8347	16.0	480	1.0960	0.8963	0.8972	0.8963	6.8233	16	1	11.0258	1.4724
0.8166	17.0	510	1.1009	0.8973	0.8974	0.8969	6.7914	16	2	11.0135	1.227
0.8054	18.0	540	1.1015	0.8957	0.8972	0.896	6.8896	17	1	11.0871	1.9632
0.786	19.0	570	1.1064	0.896	0.897	0.8961	6.8356	16	2	11.038	1.7178
0.7764	20.0	600	1.1000	0.8964	0.8965	0.896	6.7951	16	3	10.9804	1.5951
0.7526	21.0	630	1.1040	0.8961	0.8976	0.8964	6.8663	17	3	11.0748	1.7178
0.7467	22.0	660	1.1051	0.8953	0.8964	0.8954	6.8184	16	3	11.0221	1.5951
0.734	23.0	690	1.1057	0.8965	0.897	0.8963	6.8307	16	2	11.0049	1.5951
0.7268	24.0	720	1.1027	0.8956	0.8973	0.896	6.9301	17	3	11.1153	1.8405
0.718	25.0	750	1.1062	0.8965	0.8971	0.8963	6.8258	16	2	11.016	1.5951
0.7068	26.0	780	1.1058	0.8961	0.8967	0.896	6.816	16	2	11.0061	1.4724
0.6985	27.0	810	1.1120	0.8961	0.8977	0.8965	6.8933	16	2	11.1018	1.9632
0.6831	28.0	840	1.1130	0.8965	0.8968	0.8962	6.8184	16	2	11.0037	1.7178
0.6769	29.0	870	1.1144	0.8973	0.8975	0.897	6.7779	17	2	10.989	1.4724
0.6803	30.0	900	1.1139	0.8976	0.898	0.8974	6.8098	17	2	10.9779	1.5951
0.6618	31.0	930	1.1147	0.8973	0.8978	0.8971	6.8037	17	2	10.9902	1.227
0.6745	32.0	960	1.1157	0.8962	0.897	0.8961	6.8307	16	2	11.0135	1.4724
0.6618	33.0	990	1.1193	0.8963	0.897	0.8962	6.8123	17	2	10.9951	1.3497
0.6572	34.0	1020	1.1223	0.897	0.8977	0.8969	6.8209	16	2	11.0037	1.4724
0.6562	35.0	1050	1.1240	0.8963	0.8971	0.8963	6.854	17	2	11.0196	1.7178
0.6433	36.0	1080	1.1233	0.8969	0.8967	0.8964	6.8049	16	2	10.9632	1.4724
0.6405	37.0	1110	1.1236	0.8974	0.8977	0.8971	6.8245	16	2	11.011	1.5951
0.645	38.0	1140	1.1239	0.8967	0.897	0.8964	6.8135	16	2	10.9902	1.8405
0.6409	39.0	1170	1.1244	0.8967	0.897	0.8964	6.8086	16	2	10.9939	1.5951
0.6371	40.0	1200	1.1244	0.8967	0.8969	0.8964	6.8061	16	2	10.9902	1.5951

Framework versions

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