text_shortening_model_v50

This model is a fine-tuned version of facebook/bart-large-xsum on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.8296
• Rouge1: 0.5063
• Rouge2: 0.2803
• Rougel: 0.4415
• Rougelsum: 0.4405
• Bert precision: 0.8741
• Bert recall: 0.8787
• Average word count: 8.7857
• Max word count: 16
• Min word count: 3
• Average token count: 16.3942
• % shortened texts with length > 12: 11.9048

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Bert precision	Bert recall	Average word count	Max word count	Min word count	Average token count	% shortened texts with length > 12
1.3343	1.0	83	1.4625	0.5101	0.2866	0.4536	0.4527	0.8751	0.877	8.3042	19	4	15.1508	5.0265
0.7011	2.0	166	1.4296	0.5101	0.284	0.4548	0.4551	0.8736	0.8797	8.7593	18	5	16.0529	7.672
0.483	3.0	249	1.3880	0.5025	0.2819	0.4433	0.442	0.8722	0.8782	8.7698	18	5	14.8492	6.3492
0.3876	4.0	332	1.7614	0.4934	0.2653	0.4334	0.4327	0.8715	0.8725	8.2249	18	5	16.3042	5.5556
0.291	5.0	415	1.8296	0.5063	0.2803	0.4415	0.4405	0.8741	0.8787	8.7857	16	3	16.3942	11.9048

Framework versions

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