text_shortening_model_v68

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

• Loss: 1.1011
• Bert precision: 0.8904
• Bert recall: 0.8919
• Bert f1-score: 0.8906
• Average word count: 6.7117
• Max word count: 18
• Min word count: 2
• Average token count: 10.7497
• % 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: 5e-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
2.3576	1.0	37	1.6358	0.847	0.8396	0.8424	6.3233	16	0	10.7958	6.3063
1.7631	2.0	74	1.4404	0.8739	0.8715	0.8721	6.5666	16	0	10.4695	2.3023
1.5884	3.0	111	1.3492	0.8807	0.8787	0.8791	6.6006	16	2	10.3904	2.002
1.5068	4.0	148	1.2953	0.8829	0.881	0.8813	6.6006	16	2	10.4194	1.7017
1.4361	5.0	185	1.2638	0.8847	0.8836	0.8836	6.6547	16	2	10.4965	1.3013
1.3844	6.0	222	1.2357	0.8851	0.8846	0.8843	6.6747	16	2	10.5105	1.9019
1.351	7.0	259	1.2146	0.8832	0.8858	0.8839	6.8649	16	2	10.7628	2.3023
1.2944	8.0	296	1.2008	0.8848	0.8867	0.8852	6.7728	15	2	10.7047	2.1021
1.2785	9.0	333	1.1889	0.8856	0.8872	0.8858	6.7538	16	2	10.6987	1.8018
1.2469	10.0	370	1.1774	0.8851	0.8868	0.8854	6.7247	15	2	10.6627	2.1021
1.2206	11.0	407	1.1674	0.886	0.8882	0.8865	6.7558	16	2	10.7477	1.9019
1.1955	12.0	444	1.1614	0.8851	0.8875	0.8858	6.7748	15	2	10.7848	1.9019
1.1707	13.0	481	1.1516	0.8854	0.8879	0.8861	6.7698	15	2	10.7908	2.002
1.1455	14.0	518	1.1470	0.8871	0.8882	0.8872	6.6817	17	1	10.6867	1.9019
1.1392	15.0	555	1.1384	0.8861	0.8889	0.887	6.7658	17	1	10.8008	1.8018
1.1212	16.0	592	1.1351	0.8876	0.8902	0.8883	6.7528	17	1	10.8078	2.002
1.0965	17.0	629	1.1316	0.8861	0.8893	0.8872	6.7918	17	1	10.8639	2.3023
1.1	18.0	666	1.1269	0.8869	0.8901	0.8879	6.8218	17	2	10.8809	2.2022
1.0679	19.0	703	1.1220	0.8867	0.8889	0.8873	6.7157	17	1	10.7658	1.5015
1.0708	20.0	740	1.1209	0.8865	0.8889	0.8872	6.7618	17	1	10.7898	1.8018
1.0444	21.0	777	1.1178	0.8872	0.8892	0.8877	6.7047	17	2	10.7598	1.8018
1.0347	22.0	814	1.1161	0.8882	0.8902	0.8887	6.7167	17	2	10.7568	1.6016
1.0212	23.0	851	1.1147	0.8883	0.89	0.8886	6.7017	17	2	10.7467	1.8018
1.0264	24.0	888	1.1113	0.8879	0.8899	0.8884	6.6987	17	2	10.7397	1.8018
1.0186	25.0	925	1.1099	0.8876	0.8893	0.8879	6.6997	17	2	10.7417	1.7017
1.0124	26.0	962	1.1102	0.8882	0.8903	0.8888	6.7277	17	2	10.7718	2.1021
1.0081	27.0	999	1.1082	0.8889	0.8901	0.889	6.6687	17	2	10.6877	1.7017
1.0107	28.0	1036	1.1044	0.8893	0.8906	0.8894	6.6567	17	2	10.6807	1.7017
0.9788	29.0	1073	1.1060	0.8891	0.8905	0.8893	6.6817	18	2	10.7137	2.002
0.9899	30.0	1110	1.1052	0.8894	0.8915	0.8899	6.7357	18	2	10.7598	2.2022
0.9736	31.0	1147	1.1050	0.8896	0.8915	0.8901	6.7027	18	2	10.7367	2.002
0.9779	32.0	1184	1.1051	0.8899	0.892	0.8905	6.7237	18	2	10.7618	2.1021
0.9704	33.0	1221	1.1033	0.89	0.8914	0.8902	6.6877	18	2	10.7117	1.8018
0.9711	34.0	1258	1.1021	0.8894	0.8912	0.8898	6.7027	18	2	10.7327	1.8018
0.9637	35.0	1295	1.1019	0.89	0.8913	0.8901	6.6907	18	2	10.7217	1.9019
0.9525	36.0	1332	1.1016	0.8901	0.8915	0.8903	6.6997	18	2	10.7177	1.9019
0.9668	37.0	1369	1.1009	0.8902	0.8918	0.8905	6.7127	18	2	10.7497	2.002
0.9704	38.0	1406	1.1013	0.8904	0.8921	0.8908	6.7187	18	2	10.7528	2.1021
0.9531	39.0	1443	1.1010	0.8904	0.8919	0.8906	6.7117	18	2	10.7497	2.002
0.958	40.0	1480	1.1011	0.8904	0.8919	0.8906	6.7117	18	2	10.7497	2.002

Framework versions

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