text_shortening_model_v75

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

• Loss: 1.2113
• Bert precision: 0.8889
• Bert recall: 0.8883
• Bert f1-score: 0.8881
• Average word count: 6.8466
• Max word count: 15
• Min word count: 1
• Average token count: 10.892
• % shortened texts with length > 12: 1.9632

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: 1e-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
2.4857	1.0	30	1.9604	0.8298	0.8444	0.8359	9.1436	19	1	13.7337	14.2331
2.1772	2.0	60	1.7312	0.8337	0.839	0.8349	8.1264	19	1	12.3264	10.5521
1.9897	3.0	90	1.6036	0.8513	0.8528	0.8508	7.6528	19	1	11.8748	8.3436
1.8748	4.0	120	1.5274	0.8616	0.8583	0.8589	7.1988	17	1	11.4368	6.0123
1.7948	5.0	150	1.4678	0.8709	0.8669	0.868	7.0086	17	1	11.1914	4.4172
1.7436	6.0	180	1.4245	0.8763	0.8726	0.8737	6.9681	16	1	11.1387	3.8037
1.6914	7.0	210	1.3948	0.8808	0.8792	0.8793	6.9706	18	1	11.0773	3.9264
1.6484	8.0	240	1.3716	0.8846	0.8814	0.8824	6.789	15	2	10.8687	2.9448
1.6177	9.0	270	1.3534	0.8858	0.8827	0.8836	6.8294	16	2	10.8712	3.0675
1.6034	10.0	300	1.3371	0.8854	0.8826	0.8834	6.8528	16	2	10.865	2.9448
1.5696	11.0	330	1.3237	0.8863	0.8842	0.8847	6.8393	16	2	10.8577	2.6994
1.5474	12.0	360	1.3115	0.8874	0.8844	0.8853	6.7669	16	2	10.7742	2.5767
1.5354	13.0	390	1.3011	0.8867	0.8836	0.8846	6.7607	16	2	10.7644	2.3313
1.5173	14.0	420	1.2916	0.8872	0.8834	0.8847	6.7067	16	2	10.7117	2.0859
1.5061	15.0	450	1.2822	0.8873	0.8833	0.8848	6.6969	16	2	10.6945	1.9632
1.4861	16.0	480	1.2742	0.8882	0.8846	0.8858	6.692	16	2	10.7043	1.5951
1.4793	17.0	510	1.2673	0.8881	0.8848	0.8859	6.719	16	1	10.7325	1.9632
1.4736	18.0	540	1.2621	0.8888	0.8856	0.8867	6.7399	16	1	10.7571	1.9632
1.4592	19.0	570	1.2563	0.8889	0.8863	0.8871	6.7497	16	1	10.7755	1.9632
1.459	20.0	600	1.2514	0.8885	0.8863	0.8868	6.773	16	1	10.7902	1.9632
1.4446	21.0	630	1.2472	0.8883	0.8859	0.8865	6.7571	16	1	10.7546	1.8405
1.4324	22.0	660	1.2431	0.888	0.8864	0.8866	6.7779	16	1	10.7853	1.8405
1.431	23.0	690	1.2396	0.8881	0.8866	0.8868	6.7828	16	1	10.8098	1.8405
1.4233	24.0	720	1.2358	0.8885	0.8869	0.8872	6.784	16	1	10.8123	1.9632
1.4218	25.0	750	1.2322	0.8887	0.8874	0.8875	6.8135	16	1	10.8417	1.8405
1.4086	26.0	780	1.2295	0.8885	0.8878	0.8876	6.8356	16	1	10.8982	1.9632
1.4104	27.0	810	1.2267	0.8883	0.8877	0.8875	6.8491	16	1	10.9166	1.9632
1.4046	28.0	840	1.2242	0.888	0.8877	0.8873	6.8577	16	1	10.9411	1.9632
1.4034	29.0	870	1.2222	0.8882	0.8881	0.8876	6.8626	16	1	10.9436	1.9632
1.3942	30.0	900	1.2204	0.8883	0.8881	0.8877	6.8577	16	1	10.935	2.0859
1.3909	31.0	930	1.2182	0.8885	0.8881	0.8878	6.8368	15	1	10.908	1.8405
1.385	32.0	960	1.2167	0.8889	0.8884	0.8882	6.838	15	1	10.9006	1.8405
1.3833	33.0	990	1.2149	0.889	0.8884	0.8882	6.8368	15	1	10.8945	1.8405
1.3831	34.0	1020	1.2139	0.8891	0.8885	0.8883	6.8454	15	1	10.9018	1.8405
1.3811	35.0	1050	1.2129	0.8891	0.8884	0.8882	6.8356	15	1	10.8908	1.8405
1.3869	36.0	1080	1.2124	0.8891	0.8883	0.8881	6.8294	15	1	10.8785	1.8405
1.3696	37.0	1110	1.2120	0.889	0.8881	0.8881	6.8233	15	1	10.8663	1.8405
1.3791	38.0	1140	1.2116	0.8889	0.8881	0.888	6.8307	15	1	10.8748	1.8405
1.3755	39.0	1170	1.2113	0.8889	0.8881	0.888	6.8331	15	1	10.8773	1.8405
1.3668	40.0	1200	1.2113	0.8889	0.8883	0.8881	6.8466	15	1	10.892	1.9632

Framework versions

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