text_shortening_model_v80

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

• Loss: 1.1772
• Bert precision: 0.8996
• Bert recall: 0.9009
• Bert f1-score: 0.8998
• Average word count: 6.8393
• Max word count: 16
• Min word count: 3
• Average token count: 11.092
• % shortened texts with length > 12: 0.9816

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: 25

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
1.3549	1.0	30	1.0184	0.8861	0.887	0.886	7.016	18	2	11.2061	2.6994
0.9772	2.0	60	0.9395	0.889	0.8903	0.8892	6.9436	16	2	11.1276	1.8405
0.8398	3.0	90	0.9211	0.8904	0.8916	0.8906	6.9534	16	2	11.119	2.3313
0.7412	4.0	120	0.9235	0.8926	0.8945	0.8931	6.9239	16	2	11.1926	1.5951
0.6652	5.0	150	0.9173	0.8936	0.8968	0.8947	7.0442	16	3	11.4135	1.5951
0.5992	6.0	180	0.9270	0.8962	0.8982	0.8968	6.9485	16	3	11.2209	1.8405
0.5381	7.0	210	0.9565	0.8948	0.8962	0.8951	6.8209	16	2	11.1043	1.3497
0.4899	8.0	240	0.9812	0.8956	0.8984	0.8966	7.0098	16	2	11.2282	1.9632
0.4528	9.0	270	0.9842	0.8954	0.8979	0.8962	6.9791	16	3	11.2773	1.7178
0.4233	10.0	300	1.0057	0.897	0.8977	0.8969	6.8294	16	2	11.0589	1.5951
0.3971	11.0	330	1.0276	0.8967	0.8976	0.8967	6.8761	16	2	11.1411	1.1043
0.3713	12.0	360	1.0316	0.8962	0.8958	0.8955	6.7583	16	2	10.9816	1.1043
0.3428	13.0	390	1.0775	0.898	0.8982	0.8977	6.838	16	2	11.092	1.1043
0.3256	14.0	420	1.0831	0.8987	0.8993	0.8985	6.8552	16	2	11.1141	1.227
0.3116	15.0	450	1.0982	0.8979	0.899	0.898	6.8638	16	2	11.119	1.1043
0.2958	16.0	480	1.1273	0.8965	0.8991	0.8974	6.9546	16	3	11.238	1.5951
0.2838	17.0	510	1.1205	0.8984	0.9003	0.8989	6.9583	16	3	11.227	1.4724
0.2683	18.0	540	1.1435	0.8978	0.8991	0.898	6.8847	16	2	11.1178	1.227
0.2594	19.0	570	1.1495	0.899	0.8986	0.8983	6.7939	16	2	11.0307	0.8589
0.2522	20.0	600	1.1621	0.8993	0.8992	0.8988	6.7767	16	3	11.0294	0.7362
0.2457	21.0	630	1.1693	0.8991	0.9017	0.9	6.9006	16	3	11.2	0.9816
0.2442	22.0	660	1.1728	0.8986	0.9008	0.8992	6.8773	16	3	11.1644	0.9816
0.235	23.0	690	1.1740	0.8986	0.9002	0.899	6.8564	16	3	11.1178	0.9816
0.2319	24.0	720	1.1751	0.8995	0.9008	0.8997	6.8417	16	3	11.0908	0.9816
0.2315	25.0	750	1.1772	0.8996	0.9009	0.8998	6.8393	16	3	11.092	0.9816

Framework versions

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