text_shortening_model_v73

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

• Loss: 1.6126
• Bert precision: 0.9015
• Bert recall: 0.9014
• Bert f1-score: 0.901
• Average word count: 6.4004
• Max word count: 16
• Min word count: 2
• Average token count: 10.4705
• % shortened texts with length > 12: 1.1011

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: 0.0005
• 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
1.5857	1.0	37	1.1932	0.8846	0.8848	0.8842	6.5315	16	1	10.4525	1.7017
1.184	2.0	74	1.0965	0.8918	0.8915	0.8911	6.4855	17	2	10.4735	0.5005
1.0114	3.0	111	1.0773	0.8895	0.8962	0.8924	6.8959	18	2	10.995	1.3013
0.8887	4.0	148	1.0798	0.8947	0.8936	0.8937	6.4454	17	2	10.4605	1.8018
0.7851	5.0	185	1.0807	0.8941	0.8948	0.894	6.5676	16	2	10.6016	1.6016
0.7116	6.0	222	1.1002	0.8984	0.8978	0.8976	6.4605	15	2	10.4174	1.2012
0.6472	7.0	259	1.1171	0.8982	0.8997	0.8985	6.5836	16	2	10.6426	1.3013
0.5872	8.0	296	1.1196	0.8998	0.9015	0.9002	6.5415	16	2	10.6226	1.5015
0.5393	9.0	333	1.1739	0.9007	0.8979	0.8988	6.3333	16	2	10.3063	1.1011
0.4879	10.0	370	1.2079	0.8997	0.8983	0.8985	6.3343	15	2	10.2913	1.001
0.4615	11.0	407	1.2230	0.8988	0.8997	0.8988	6.5165	15	2	10.6426	1.3013
0.4245	12.0	444	1.2325	0.8996	0.8979	0.8983	6.3704	15	2	10.4334	1.3013
0.3973	13.0	481	1.2657	0.8973	0.8987	0.8975	6.4855	15	2	10.5876	1.6016
0.3658	14.0	518	1.2875	0.8985	0.8993	0.8984	6.4735	15	2	10.5355	1.2012
0.3422	15.0	555	1.3202	0.9002	0.8991	0.8992	6.2873	14	2	10.3594	1.001
0.3271	16.0	592	1.3315	0.9006	0.9	0.8998	6.3784	15	2	10.4454	0.9009
0.305	17.0	629	1.3441	0.8994	0.9005	0.8995	6.4705	16	2	10.5906	1.2012
0.2847	18.0	666	1.3648	0.8997	0.8989	0.8989	6.3584	14	2	10.4244	0.9009
0.2707	19.0	703	1.3837	0.9005	0.9011	0.9003	6.4545	16	2	10.5365	1.3013
0.254	20.0	740	1.4180	0.8997	0.9006	0.8997	6.4444	15	2	10.5516	1.2012
0.2421	21.0	777	1.4100	0.9014	0.903	0.9017	6.4755	16	2	10.6016	0.9009
0.2301	22.0	814	1.4437	0.9	0.901	0.9	6.4825	15	2	10.5626	0.8008
0.2183	23.0	851	1.4762	0.9003	0.9014	0.9004	6.4995	16	2	10.6116	1.3013
0.2148	24.0	888	1.4815	0.9007	0.9014	0.9006	6.4484	16	2	10.5495	1.1011
0.2013	25.0	925	1.5039	0.9018	0.9015	0.9012	6.4144	15	2	10.4925	1.001
0.1924	26.0	962	1.5217	0.9013	0.9014	0.9009	6.4024	16	2	10.4765	1.2012
0.1854	27.0	999	1.5125	0.902	0.9014	0.9012	6.3774	16	2	10.4565	1.1011
0.1769	28.0	1036	1.5384	0.8998	0.9011	0.9	6.4925	16	2	10.6106	1.001
0.1713	29.0	1073	1.5627	0.9012	0.9018	0.901	6.4715	16	2	10.5395	1.2012
0.1685	30.0	1110	1.5473	0.9011	0.9004	0.9002	6.4064	16	2	10.4484	1.1011
0.1681	31.0	1147	1.5592	0.9018	0.9018	0.9013	6.4194	15	2	10.5165	0.8008
0.1599	32.0	1184	1.5800	0.9006	0.9007	0.9002	6.4254	16	2	10.5005	1.001
0.1509	33.0	1221	1.5822	0.9012	0.9005	0.9004	6.3994	16	2	10.4314	1.001
0.1509	34.0	1258	1.5924	0.9013	0.9008	0.9006	6.4084	16	2	10.4655	1.1011
0.1408	35.0	1295	1.6045	0.9028	0.9024	0.9021	6.4074	16	2	10.4845	1.2012
0.1487	36.0	1332	1.6133	0.9014	0.9012	0.9008	6.4244	16	2	10.4775	1.001
0.1444	37.0	1369	1.6157	0.9016	0.9016	0.9012	6.4304	16	2	10.5045	1.2012
0.1418	38.0	1406	1.6105	0.9012	0.9011	0.9006	6.4084	16	2	10.4615	1.1011
0.1402	39.0	1443	1.6116	0.9017	0.9015	0.9011	6.3894	16	2	10.4494	1.1011
0.1375	40.0	1480	1.6126	0.9015	0.9014	0.901	6.4004	16	2	10.4705	1.1011

Framework versions

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