text_shortening_model_v65

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

• Loss: 1.1783
• Bert precision: 0.8964
• Bert recall: 0.8977
• Bert f1-score: 0.8966
• Average word count: 6.4565
• Max word count: 16
• Min word count: 2
• Average token count: 10.5686
• % shortened texts with length > 12: 2.002

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30

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
1.7747	1.0	146	1.3200	0.8806	0.8825	0.881	6.7818	18	2	10.6827	2.1021
1.3684	2.0	292	1.2106	0.8857	0.8858	0.8852	6.5335	18	2	10.4835	1.7017
1.2448	3.0	438	1.1635	0.8862	0.8883	0.8868	6.6246	18	1	10.6817	2.1021
1.1406	4.0	584	1.1386	0.8897	0.8923	0.8905	6.6697	18	2	10.6767	2.2022
1.0623	5.0	730	1.1373	0.889	0.893	0.8905	6.6897	18	2	10.7568	1.5015
1.0034	6.0	876	1.1111	0.8923	0.8953	0.8933	6.5876	18	2	10.6927	1.7017
0.9391	7.0	1022	1.1037	0.8927	0.8947	0.8932	6.5455	18	2	10.6196	1.3013
0.8868	8.0	1168	1.0997	0.8949	0.8959	0.895	6.4805	18	2	10.5836	1.4014
0.8443	9.0	1314	1.1011	0.8939	0.8965	0.8947	6.5626	18	2	10.6386	1.5015
0.8117	10.0	1460	1.0997	0.8957	0.8981	0.8965	6.4865	16	2	10.6066	1.001
0.7844	11.0	1606	1.1153	0.8976	0.8979	0.8973	6.4404	18	2	10.5345	1.5015
0.7593	12.0	1752	1.1126	0.8946	0.8988	0.8962	6.6356	18	2	10.7698	1.9019
0.7249	13.0	1898	1.1047	0.8968	0.8991	0.8975	6.5335	16	2	10.6396	1.4014
0.7048	14.0	2044	1.1127	0.8961	0.8984	0.8968	6.5275	16	2	10.6336	1.4014
0.6828	15.0	2190	1.1237	0.8965	0.8982	0.8969	6.4675	16	2	10.5906	1.7017
0.6558	16.0	2336	1.1221	0.8975	0.8972	0.8969	6.3634	16	1	10.4985	1.2012
0.6296	17.0	2482	1.1296	0.8962	0.8982	0.8968	6.4775	16	1	10.6496	1.9019
0.6304	18.0	2628	1.1334	0.8981	0.898	0.8976	6.3724	16	1	10.4755	1.6016
0.6124	19.0	2774	1.1463	0.898	0.9006	0.8989	6.5075	15	2	10.6246	1.5015
0.6001	20.0	2920	1.1547	0.8982	0.8997	0.8984	6.4925	16	2	10.5766	1.9019
0.5834	21.0	3066	1.1551	0.8972	0.8973	0.8967	6.3323	16	2	10.4705	1.7017
0.5707	22.0	3212	1.1687	0.897	0.899	0.8976	6.4665	16	2	10.6026	1.7017
0.5667	23.0	3358	1.1656	0.8965	0.8981	0.8968	6.4585	16	2	10.5726	2.002
0.5519	24.0	3504	1.1747	0.8968	0.8984	0.8971	6.4885	16	2	10.5616	2.1021
0.5538	25.0	3650	1.1754	0.8967	0.8983	0.897	6.4735	16	2	10.5676	2.002
0.5403	26.0	3796	1.1734	0.8968	0.8983	0.8971	6.4835	16	2	10.6036	1.9019
0.5371	27.0	3942	1.1735	0.8964	0.8982	0.8968	6.4865	16	2	10.5696	2.1021
0.5381	28.0	4088	1.1767	0.8968	0.8982	0.897	6.4735	16	2	10.5926	1.9019
0.5278	29.0	4234	1.1771	0.8966	0.8975	0.8966	6.4454	16	2	10.5556	2.002
0.5249	30.0	4380	1.1783	0.8964	0.8977	0.8966	6.4565	16	2	10.5686	2.002

Framework versions

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