<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
text_shortening_model_v39
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: 2.8730
- Rouge1: 0.4929
- Rouge2: 0.2546
- Rougel: 0.4351
- Rougelsum: 0.4353
- Bert precision: 0.8698
- Bert recall: 0.8762
- Average word count: 8.8348
- Max word count: 17
- Min word count: 4
- Average token count: 16.5796
- % shortened texts with length > 12: 8.4084
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: 20
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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.9582 | 1.0 | 73 | 1.4062 | 0.5229 | 0.2983 | 0.4739 | 0.4738 | 0.875 | 0.8853 | 8.9039 | 17 | 4 | 15.0811 | 9.009 |
| 0.5598 | 2.0 | 146 | 1.4819 | 0.5053 | 0.2806 | 0.456 | 0.4561 | 0.8723 | 0.879 | 8.6486 | 14 | 5 | 14.2703 | 1.5015 |
| 0.3791 | 3.0 | 219 | 1.7718 | 0.5174 | 0.2882 | 0.4532 | 0.4539 | 0.8705 | 0.8834 | 9.6456 | 18 | 5 | 17.7027 | 16.5165 |
| 0.3748 | 4.0 | 292 | 2.1513 | 0.3078 | 0.1184 | 0.2773 | 0.278 | 0.8215 | 0.8336 | 9.5375 | 18 | 4 | 17.1441 | 9.9099 |
| 0.2837 | 5.0 | 365 | 1.6757 | 0.4999 | 0.2661 | 0.4487 | 0.4489 | 0.8732 | 0.8766 | 8.3844 | 16 | 4 | 15.1892 | 6.6066 |
| 0.1885 | 6.0 | 438 | 1.8005 | 0.4938 | 0.2619 | 0.4437 | 0.4439 | 0.8729 | 0.8763 | 8.5526 | 14 | 5 | 14.994 | 1.5015 |
| 0.1799 | 7.0 | 511 | 1.8427 | 0.4986 | 0.2752 | 0.4455 | 0.4463 | 0.8664 | 0.8796 | 9.4384 | 20 | 5 | 15.6697 | 11.4114 |
| 0.1638 | 8.0 | 584 | 2.0234 | 0.5206 | 0.2854 | 0.4632 | 0.4642 | 0.8774 | 0.8844 | 9.1682 | 18 | 4 | 16.2132 | 9.9099 |
| 0.1247 | 9.0 | 657 | 1.9158 | 0.486 | 0.2628 | 0.4326 | 0.4339 | 0.8707 | 0.8758 | 8.7327 | 17 | 4 | 15.3093 | 6.6066 |
| 0.1059 | 10.0 | 730 | 2.2355 | 0.5127 | 0.2825 | 0.4578 | 0.4577 | 0.875 | 0.8827 | 9.045 | 17 | 4 | 16.5586 | 8.7087 |
| 0.1104 | 11.0 | 803 | 2.2555 | 0.5095 | 0.2698 | 0.4514 | 0.4511 | 0.8762 | 0.8815 | 8.7928 | 17 | 4 | 16.3123 | 8.7087 |
| 0.1196 | 12.0 | 876 | 2.3329 | 0.507 | 0.2692 | 0.453 | 0.454 | 0.8746 | 0.8795 | 8.8228 | 15 | 5 | 16.1862 | 5.4054 |
| 0.093 | 13.0 | 949 | 2.2657 | 0.5137 | 0.2748 | 0.4545 | 0.4543 | 0.8733 | 0.8801 | 8.7988 | 16 | 4 | 16.012 | 7.8078 |
| 0.0626 | 14.0 | 1022 | 2.5004 | 0.5014 | 0.2677 | 0.4432 | 0.4435 | 0.8725 | 0.8775 | 8.7508 | 16 | 5 | 16.4535 | 6.9069 |
| 0.0534 | 15.0 | 1095 | 2.4192 | 0.5031 | 0.27 | 0.4467 | 0.447 | 0.8711 | 0.8784 | 8.8438 | 19 | 4 | 16.1411 | 9.3093 |
| 0.0475 | 16.0 | 1168 | 2.5800 | 0.4891 | 0.2553 | 0.4313 | 0.4315 | 0.8689 | 0.8753 | 8.8408 | 18 | 4 | 16.5045 | 8.7087 |
| 0.0399 | 17.0 | 1241 | 2.6858 | 0.5021 | 0.2615 | 0.4452 | 0.445 | 0.8727 | 0.8782 | 8.7808 | 17 | 4 | 16.3844 | 7.2072 |
| 0.0296 | 18.0 | 1314 | 2.6646 | 0.4992 | 0.2666 | 0.4466 | 0.4463 | 0.8726 | 0.8764 | 8.5706 | 17 | 4 | 16.1111 | 4.8048 |
| 0.0286 | 19.0 | 1387 | 2.7496 | 0.5023 | 0.2648 | 0.4451 | 0.445 | 0.8721 | 0.8781 | 8.7868 | 17 | 4 | 16.3063 | 6.6066 |
| 0.026 | 20.0 | 1460 | 2.8730 | 0.4929 | 0.2546 | 0.4351 | 0.4353 | 0.8698 | 0.8762 | 8.8348 | 17 | 4 | 16.5796 | 8.4084 |
Framework versions
- Transformers 4.33.1
- Pytorch 2.0.1+cu118
- Datasets 2.14.5
- Tokenizers 0.13.3
