text_shortening_model_v35

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: 2.1783
• Rouge1: 0.4993
• Rouge2: 0.2724
• Rougel: 0.4472
• Rougelsum: 0.4467
• Bert precision: 0.8744
• Bert recall: 0.8769
• Average word count: 8.6096
• Max word count: 20
• Min word count: 5
• Average token count: 14.97
• % shortened texts with length > 12: 3.6036

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: 0.0003
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

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
1.83	1.0	37	1.9642	0.457	0.2329	0.4049	0.4054	0.8677	0.8663	8.027	13	4	16.6607	2.7027
0.8629	2.0	74	1.6943	0.5268	0.3019	0.4695	0.4697	0.8758	0.8901	10.0571	19	5	17.8258	19.2192
0.7849	3.0	111	1.6564	0.5001	0.279	0.4553	0.4554	0.873	0.8805	8.9099	17	5	15.4865	5.1051
0.6116	4.0	148	1.7559	0.4638	0.2376	0.4183	0.4188	0.863	0.8665	8.4414	15	4	13.8829	0.9009
0.3976	5.0	185	1.6708	0.4999	0.2723	0.4481	0.4481	0.8744	0.8766	8.5556	16	5	14.6877	3.9039
0.2977	6.0	222	1.7196	0.4937	0.2699	0.4376	0.4379	0.8684	0.877	9.1652	20	5	15.3964	5.7057
0.2187	7.0	259	1.7942	0.5129	0.2905	0.4572	0.4575	0.8765	0.8803	8.7117	19	5	14.6306	3.9039
0.1603	8.0	296	1.8003	0.4822	0.2538	0.4237	0.4229	0.8688	0.8722	8.6306	19	5	15.4474	5.7057
0.1175	9.0	333	2.0138	0.5024	0.2798	0.4486	0.4475	0.8742	0.8791	8.7988	19	5	16.1471	6.6066
0.0859	10.0	370	2.1783	0.4993	0.2724	0.4472	0.4467	0.8744	0.8769	8.6096	20	5	14.97	3.6036

Framework versions

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