Edit model card

text_shortening_model_v43

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: 2.8362
  • Rouge1: 0.4977
  • Rouge2: 0.2645
  • Rougel: 0.4429
  • Rougelsum: 0.4422
  • Bert precision: 0.8744
  • Bert recall: 0.8788
  • Average word count: 8.5344
  • Max word count: 18
  • Min word count: 4
  • Average token count: 15.9365
  • % shortened texts with length > 12: 8.4656

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
0.5902 1.0 83 1.5909 0.4855 0.2475 0.4202 0.4201 0.8682 0.8736 8.5661 15 4 16.0635 3.9683
0.383 2.0 166 1.4957 0.516 0.2977 0.4569 0.4567 0.8751 0.881 8.8016 17 4 16.3519 8.4656
0.3301 3.0 249 1.6999 0.5073 0.2678 0.4401 0.4402 0.8662 0.8856 10.4233 22 5 17.9286 24.6032
0.3264 4.0 332 1.5703 0.5121 0.2818 0.4525 0.4527 0.8716 0.8844 9.1561 19 4 15.8704 12.4339
0.3901 5.0 415 1.6559 0.4875 0.2629 0.4362 0.4365 0.8661 0.8772 9.1111 16 5 15.2275 5.0265
0.2982 6.0 498 1.8927 0.499 0.267 0.4479 0.4476 0.8724 0.8824 9.0185 17 5 16.6376 10.0529
0.2864 7.0 581 1.8092 0.4961 0.2673 0.4377 0.4372 0.8705 0.8789 8.6614 17 5 14.4656 5.291
0.2059 8.0 664 2.0127 0.4921 0.2652 0.4408 0.4408 0.8729 0.8778 8.5899 16 4 15.2725 6.8783
0.1655 9.0 747 2.1199 0.4886 0.2697 0.4392 0.4391 0.8713 0.8777 8.7011 16 4 16.0132 7.4074
0.2361 10.0 830 2.0002 0.4814 0.2536 0.427 0.4257 0.8666 0.8769 8.9921 19 4 15.037 6.0847
0.2329 11.0 913 2.3033 0.4961 0.2725 0.4441 0.4426 0.8722 0.8775 8.6958 17 5 16.2619 10.582
0.1743 12.0 996 2.4562 0.499 0.275 0.4474 0.4477 0.8745 0.878 8.4127 17 4 15.873 9.2593
0.1716 13.0 1079 2.4160 0.4811 0.2528 0.4299 0.4297 0.8708 0.8751 8.4735 16 4 16.0873 6.0847
0.1394 14.0 1162 2.3996 0.4783 0.2445 0.4214 0.4205 0.8686 0.8735 8.6587 19 5 15.6376 8.9947
0.0769 15.0 1245 2.8364 0.4902 0.258 0.4369 0.4362 0.8697 0.8767 8.7222 18 4 16.4286 9.5238
0.1039 16.0 1328 2.5845 0.5009 0.267 0.4473 0.4464 0.8757 0.88 8.5291 18 4 16.0688 8.7302
0.098 17.0 1411 2.7602 0.491 0.2628 0.4379 0.4377 0.8711 0.8779 8.6587 18 4 16.2249 9.7884
0.0879 18.0 1494 2.6813 0.4987 0.2679 0.4468 0.4471 0.8761 0.8793 8.3862 18 4 15.4735 7.9365
0.0945 19.0 1577 2.8612 0.5034 0.2703 0.4489 0.449 0.8762 0.8806 8.5582 19 4 16.0873 8.4656
0.0702 20.0 1660 2.8362 0.4977 0.2645 0.4429 0.4422 0.8744 0.8788 8.5344 18 4 15.9365 8.4656

Framework versions

  • Transformers 4.33.1
  • Pytorch 2.0.1+cu118
  • Datasets 2.14.5
  • Tokenizers 0.13.3
Downloads last month
1
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for ldos/text_shortening_model_v43

Finetuned
(50)
this model