text_shortening_model_v70

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

• Loss: 1.1285
• Bert precision: 0.8931
• Bert recall: 0.8981
• Bert f1-score: 0.8951
• Average word count: 6.5696
• Max word count: 16
• Min word count: 2
• Average token count: 10.6276
• % shortened texts with length > 12: 1.8018

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
2.1122	1.0	37	1.5033	0.8698	0.8713	0.87	6.6056	16	0	10.4705	3.003
1.5423	2.0	74	1.3536	0.8779	0.8809	0.8788	6.6286	16	2	10.5696	1.8018
1.4111	3.0	111	1.2926	0.8805	0.8849	0.8821	6.6446	16	2	10.5586	2.002
1.3269	4.0	148	1.2394	0.8852	0.8889	0.8865	6.5636	15	2	10.5175	1.4014
1.2566	5.0	185	1.2094	0.884	0.8889	0.8859	6.6787	16	2	10.6226	2.2022
1.1991	6.0	222	1.1820	0.8848	0.8899	0.8868	6.6747	16	2	10.6266	2.3023
1.1567	7.0	259	1.1627	0.8849	0.8916	0.8878	6.7908	16	2	10.7067	2.002
1.1126	8.0	296	1.1560	0.8866	0.8936	0.8896	6.7638	16	2	10.7598	2.1021
1.0722	9.0	333	1.1501	0.889	0.8944	0.8912	6.6807	16	2	10.6456	2.1021
1.0455	10.0	370	1.1384	0.8876	0.8946	0.8906	6.7117	16	2	10.7317	2.002
1.0111	11.0	407	1.1220	0.8907	0.8951	0.8924	6.5706	16	2	10.5425	1.5015
0.9804	12.0	444	1.1243	0.8912	0.8963	0.8933	6.5766	16	2	10.5836	1.6016
0.9509	13.0	481	1.1256	0.8898	0.8959	0.8924	6.6226	16	2	10.6366	1.9019
0.9295	14.0	518	1.1206	0.8896	0.8972	0.8929	6.7287	16	2	10.7788	2.3023
0.9137	15.0	555	1.1172	0.8917	0.895	0.8929	6.5175	16	2	10.4885	1.7017
0.9016	16.0	592	1.1218	0.8902	0.8978	0.8935	6.7237	16	2	10.7608	1.9019
0.8654	17.0	629	1.1171	0.8913	0.8966	0.8934	6.5946	16	2	10.6166	2.002
0.8562	18.0	666	1.1194	0.8916	0.8973	0.8939	6.6186	16	2	10.6607	2.002
0.8337	19.0	703	1.1235	0.8921	0.8989	0.895	6.7027	16	2	10.7658	1.8018
0.8323	20.0	740	1.1153	0.8914	0.8977	0.8941	6.6607	16	2	10.6917	1.8018
0.8146	21.0	777	1.1142	0.8929	0.8966	0.8943	6.5315	16	2	10.5536	1.7017
0.8053	22.0	814	1.1211	0.8923	0.8983	0.8948	6.6747	16	2	10.7407	2.1021
0.7858	23.0	851	1.1164	0.8928	0.8969	0.8944	6.5355	16	2	10.5706	1.6016
0.7795	24.0	888	1.1157	0.8942	0.8983	0.8958	6.5556	16	2	10.6016	1.7017
0.7603	25.0	925	1.1231	0.8935	0.8984	0.8955	6.5826	16	2	10.6486	1.7017
0.7709	26.0	962	1.1231	0.8932	0.8979	0.8951	6.6006	16	2	10.6396	2.002
0.7528	27.0	999	1.1217	0.8929	0.8973	0.8946	6.5506	16	2	10.5876	1.9019
0.7436	28.0	1036	1.1222	0.8933	0.8991	0.8957	6.5696	16	2	10.6587	2.002
0.7406	29.0	1073	1.1284	0.8928	0.898	0.8949	6.5636	16	2	10.6406	1.9019
0.7326	30.0	1110	1.1278	0.8932	0.8988	0.8956	6.6216	16	2	10.6977	1.9019
0.7253	31.0	1147	1.1273	0.893	0.8986	0.8953	6.5856	16	2	10.6537	1.7017
0.7245	32.0	1184	1.1274	0.8935	0.8985	0.8955	6.5586	16	2	10.6216	1.7017
0.7082	33.0	1221	1.1281	0.8935	0.8987	0.8956	6.5826	16	2	10.6597	1.8018
0.7011	34.0	1258	1.1265	0.8936	0.8985	0.8956	6.5696	16	2	10.6517	1.8018
0.7099	35.0	1295	1.1284	0.8935	0.8987	0.8956	6.5656	16	2	10.6396	1.6016
0.7064	36.0	1332	1.1281	0.8939	0.8992	0.896	6.5696	16	2	10.6476	1.8018
0.7045	37.0	1369	1.1286	0.8932	0.8978	0.895	6.5556	16	2	10.5976	1.8018
0.702	38.0	1406	1.1275	0.8934	0.8984	0.8954	6.5776	16	2	10.6296	1.8018
0.6952	39.0	1443	1.1284	0.8933	0.8981	0.8952	6.5586	16	2	10.6146	1.8018
0.6928	40.0	1480	1.1285	0.8931	0.8981	0.8951	6.5696	16	2	10.6276	1.8018

Framework versions

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