text_shortening_model_v74

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

• Loss: 1.2644
• Bert precision: 0.8826
• Bert recall: 0.8851
• Bert f1-score: 0.8832
• Average word count: 6.7137
• Max word count: 16
• Min word count: 2
• Average token count: 10.6547
• % shortened texts with length > 12: 2.6026

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
2.5095	1.0	37	1.9952	0.8247	0.8396	0.8308	8.5926	19	0	12.961	12.8128
2.1271	2.0	74	1.7552	0.8393	0.8454	0.841	7.7247	17	0	11.7738	9.3093
1.9629	3.0	111	1.6420	0.8552	0.8582	0.8556	7.2022	17	1	11.3193	6.5065
1.8511	4.0	148	1.5687	0.8646	0.8639	0.8634	6.8078	17	1	10.8539	4.1041
1.7806	5.0	185	1.5196	0.8684	0.8693	0.8681	6.8278	16	1	10.8438	3.8038
1.7193	6.0	222	1.4840	0.8713	0.8736	0.8717	6.8388	18	2	10.8318	3.4034
1.6763	7.0	259	1.4540	0.8756	0.8765	0.8754	6.7528	18	2	10.6847	3.003
1.6389	8.0	296	1.4316	0.8766	0.8785	0.8769	6.7628	16	2	10.6917	2.6026
1.6146	9.0	333	1.4149	0.8771	0.8798	0.8778	6.8018	15	2	10.7177	2.7027
1.597	10.0	370	1.3986	0.8782	0.8811	0.879	6.7998	15	2	10.7067	2.5025
1.5761	11.0	407	1.3860	0.8792	0.8815	0.8797	6.7588	15	2	10.6496	2.3023
1.5456	12.0	444	1.3747	0.8792	0.8813	0.8797	6.7387	16	2	10.6376	2.2022
1.533	13.0	481	1.3647	0.88	0.8823	0.8805	6.7347	16	2	10.6276	2.1021
1.5142	14.0	518	1.3536	0.8805	0.8822	0.8808	6.7047	16	2	10.5746	1.9019
1.514	15.0	555	1.3429	0.8803	0.882	0.8805	6.6847	16	2	10.5606	1.7017
1.4973	16.0	592	1.3353	0.8805	0.8828	0.881	6.7467	16	2	10.6627	2.1021
1.4792	17.0	629	1.3277	0.8811	0.8829	0.8814	6.7077	16	2	10.6166	2.002
1.4669	18.0	666	1.3206	0.8815	0.8831	0.8817	6.6927	16	2	10.6016	2.1021
1.4667	19.0	703	1.3141	0.881	0.8831	0.8815	6.7167	16	2	10.6306	2.1021
1.4497	20.0	740	1.3097	0.8808	0.883	0.8813	6.7227	16	2	10.6416	2.1021
1.4533	21.0	777	1.3053	0.8814	0.8831	0.8817	6.6997	16	2	10.6086	2.1021
1.4408	22.0	814	1.2998	0.8808	0.8825	0.881	6.7037	16	2	10.6076	2.2022
1.4343	23.0	851	1.2958	0.8807	0.8829	0.8812	6.7297	16	2	10.6306	2.3023
1.4295	24.0	888	1.2926	0.881	0.8833	0.8816	6.7427	16	2	10.6486	2.4024
1.4219	25.0	925	1.2887	0.8812	0.8835	0.8818	6.7327	16	2	10.6426	2.4024
1.4045	26.0	962	1.2855	0.8814	0.8836	0.8819	6.7187	16	2	10.6256	2.4024
1.409	27.0	999	1.2826	0.8817	0.884	0.8823	6.7217	16	2	10.6456	2.6026
1.3994	28.0	1036	1.2803	0.8826	0.8848	0.8831	6.7047	16	2	10.6226	2.7027
1.3905	29.0	1073	1.2778	0.8823	0.8847	0.8829	6.7267	16	2	10.6507	2.8028
1.4014	30.0	1110	1.2751	0.8821	0.8845	0.8827	6.7237	16	2	10.6466	2.8028
1.3946	31.0	1147	1.2732	0.8826	0.8849	0.8831	6.7167	16	2	10.6426	2.8028
1.3915	32.0	1184	1.2712	0.8823	0.8845	0.8828	6.7057	16	2	10.6336	2.7027
1.3904	33.0	1221	1.2695	0.8824	0.8847	0.883	6.7047	16	2	10.6376	2.7027
1.3843	34.0	1258	1.2684	0.8828	0.885	0.8833	6.7097	16	2	10.6406	2.6026
1.3875	35.0	1295	1.2672	0.8827	0.8852	0.8834	6.7217	16	2	10.6607	2.6026
1.3794	36.0	1332	1.2661	0.8828	0.8851	0.8834	6.7087	16	2	10.6426	2.6026
1.3906	37.0	1369	1.2654	0.8828	0.8853	0.8835	6.7177	16	2	10.6567	2.6026
1.3841	38.0	1406	1.2648	0.8826	0.8851	0.8833	6.7107	16	2	10.6476	2.6026
1.3761	39.0	1443	1.2645	0.8825	0.885	0.8832	6.7137	16	2	10.6537	2.6026
1.3797	40.0	1480	1.2644	0.8826	0.8851	0.8832	6.7137	16	2	10.6547	2.6026

Framework versions

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