Edit model card

text_shortening_model_v57

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

  • Loss: 0.8302
  • Rouge1: 0.676
  • Rouge2: 0.4816
  • Rougel: 0.6172
  • Rougelsum: 0.6179
  • Bert precision: 0.9135
  • Bert recall: 0.9082
  • Bert f1-score: 0.9104
  • Average word count: 8.1161
  • Max word count: 16
  • Min word count: 4
  • Average token count: 12.2946
  • % shortened texts with length > 12: 6.25

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: 1e-05
  • 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: 50

Training results

Training Loss Epoch Step Validation Loss Rouge1 Rouge2 Rougel Rougelsum Bert precision Bert recall Bert f1-score Average word count Max word count Min word count Average token count % shortened texts with length > 12
2.6374 1.0 49 2.0882 0.3209 0.1826 0.2886 0.2875 0.7596 0.789 0.7725 9.6562 17 0 15.9018 30.3571
2.1652 2.0 98 1.7029 0.3216 0.1848 0.2939 0.2932 0.7531 0.7906 0.77 9.7054 17 0 16.0402 25.0
1.8731 3.0 147 1.4874 0.4234 0.277 0.3972 0.3951 0.8117 0.8225 0.8158 8.7455 17 0 14.567 21.4286
1.6683 4.0 196 1.3346 0.5198 0.3525 0.4889 0.4888 0.8491 0.8338 0.8405 7.4598 17 0 12.7188 13.3929
1.5287 5.0 245 1.2321 0.5535 0.377 0.5101 0.5109 0.8713 0.8588 0.8643 7.7143 17 0 12.6384 12.0536
1.4303 6.0 294 1.1638 0.5773 0.3941 0.529 0.529 0.8848 0.8703 0.8769 7.5 17 0 12.2009 9.375
1.3696 7.0 343 1.1091 0.5945 0.4125 0.5426 0.5421 0.8946 0.8814 0.8874 7.6473 16 0 12.1875 9.375
1.305 8.0 392 1.0707 0.597 0.4082 0.5434 0.5432 0.8959 0.883 0.8889 7.6741 16 0 12.2143 8.0357
1.2575 9.0 441 1.0384 0.6094 0.4193 0.5521 0.553 0.8993 0.8868 0.8925 7.6786 17 0 12.125 7.1429
1.241 10.0 490 1.0125 0.617 0.423 0.5595 0.5601 0.9038 0.8918 0.8973 7.817 17 2 12.1027 7.5893
1.17 11.0 539 0.9912 0.6173 0.4236 0.5593 0.5591 0.9052 0.892 0.8981 7.7455 17 2 11.9911 6.25
1.1413 12.0 588 0.9750 0.6253 0.4322 0.5661 0.5663 0.9049 0.8935 0.8986 7.9286 17 2 12.1786 7.1429
1.1367 13.0 637 0.9586 0.63 0.4356 0.5704 0.5704 0.9068 0.8943 0.9 7.8705 17 2 12.0357 6.6964
1.1101 14.0 686 0.9458 0.6273 0.4355 0.5665 0.5671 0.9057 0.8949 0.8998 7.942 17 2 12.1384 6.6964
1.0711 15.0 735 0.9374 0.6357 0.4424 0.5718 0.5721 0.9068 0.8969 0.9013 8.0179 17 2 12.1875 6.6964
1.0553 16.0 784 0.9282 0.6378 0.4455 0.5752 0.5756 0.9084 0.8969 0.9022 7.8571 16 2 12.0536 4.9107
1.047 17.0 833 0.9188 0.6439 0.4525 0.5821 0.5825 0.9085 0.8996 0.9035 7.9955 16 2 12.1741 5.3571
1.0201 18.0 882 0.9104 0.643 0.4536 0.5832 0.5836 0.9083 0.8997 0.9035 8.0134 16 2 12.2098 5.8036
1.0228 19.0 931 0.9023 0.6471 0.4601 0.5862 0.5865 0.9101 0.902 0.9056 8.0 16 2 12.1652 4.9107
0.9896 20.0 980 0.8936 0.6497 0.463 0.5882 0.5888 0.9103 0.9017 0.9055 8.0491 16 2 12.1741 5.3571
0.9815 21.0 1029 0.8873 0.6555 0.4659 0.5937 0.5948 0.9106 0.9025 0.9061 8.0402 16 2 12.2411 5.8036
0.9877 22.0 1078 0.8828 0.6618 0.4728 0.6005 0.6007 0.9125 0.9047 0.9081 8.1205 16 2 12.308 6.25
0.9696 23.0 1127 0.8774 0.661 0.4679 0.6 0.5994 0.9128 0.9046 0.9082 8.0938 16 3 12.2902 6.25
0.9556 24.0 1176 0.8737 0.6613 0.4717 0.6023 0.6022 0.913 0.9052 0.9086 8.0893 16 3 12.3036 6.25
0.95 25.0 1225 0.8703 0.6636 0.4725 0.6044 0.6041 0.913 0.9055 0.9088 8.1384 16 3 12.3616 6.25
0.9464 26.0 1274 0.8660 0.6629 0.4723 0.6057 0.6052 0.9125 0.9053 0.9085 8.1562 16 3 12.3482 6.25
0.9189 27.0 1323 0.8605 0.6633 0.4746 0.6084 0.6079 0.9124 0.9052 0.9083 8.0848 16 3 12.2723 6.25
0.9277 28.0 1372 0.8583 0.662 0.4731 0.6059 0.6057 0.9118 0.9059 0.9084 8.1607 16 3 12.3304 6.25
0.9142 29.0 1421 0.8550 0.6663 0.4784 0.6106 0.6104 0.9126 0.9073 0.9095 8.1786 16 4 12.3482 6.6964
0.913 30.0 1470 0.8529 0.6656 0.477 0.6073 0.607 0.9123 0.9073 0.9093 8.2589 16 4 12.4241 7.1429
0.8984 31.0 1519 0.8507 0.6708 0.4804 0.6114 0.6116 0.9128 0.9083 0.9101 8.2098 16 4 12.3973 6.6964
0.903 32.0 1568 0.8479 0.6728 0.4777 0.6096 0.6096 0.9133 0.9081 0.9103 8.2232 16 4 12.3973 7.1429
0.8947 33.0 1617 0.8452 0.6741 0.4785 0.6119 0.6116 0.9132 0.9081 0.9101 8.1741 16 4 12.3616 7.1429
0.8883 34.0 1666 0.8424 0.6733 0.4766 0.6108 0.6107 0.9125 0.9072 0.9094 8.1607 16 4 12.3438 7.1429
0.877 35.0 1715 0.8403 0.6742 0.4799 0.6141 0.6145 0.9133 0.908 0.9102 8.1429 16 4 12.3304 6.6964
0.8612 36.0 1764 0.8393 0.6737 0.4808 0.6141 0.6143 0.9133 0.908 0.9102 8.1384 16 4 12.3259 6.6964
0.8848 37.0 1813 0.8363 0.673 0.478 0.6131 0.6133 0.9124 0.9074 0.9095 8.1384 16 4 12.3214 6.6964
0.8717 38.0 1862 0.8363 0.6729 0.478 0.613 0.6132 0.9129 0.9075 0.9097 8.0848 16 4 12.2545 5.8036
0.8739 39.0 1911 0.8355 0.6711 0.4775 0.6115 0.6118 0.913 0.9072 0.9096 8.0714 16 4 12.2366 5.8036
0.8569 40.0 1960 0.8343 0.672 0.4772 0.6125 0.6128 0.9132 0.9074 0.9098 8.0804 16 4 12.2366 5.8036
0.8601 41.0 2009 0.8342 0.675 0.4831 0.6163 0.6165 0.9139 0.9081 0.9105 8.0982 16 4 12.2634 5.8036
0.8519 42.0 2058 0.8330 0.6743 0.481 0.6147 0.6152 0.9137 0.908 0.9104 8.1027 16 4 12.2723 5.8036
0.8713 43.0 2107 0.8322 0.6757 0.4844 0.617 0.6172 0.9133 0.9079 0.9102 8.125 16 4 12.2857 6.25
0.8554 44.0 2156 0.8313 0.6746 0.4809 0.6151 0.6154 0.9132 0.9079 0.9101 8.1384 16 4 12.2857 6.25
0.8559 45.0 2205 0.8314 0.6773 0.4849 0.6184 0.6189 0.9137 0.9085 0.9106 8.125 16 4 12.2857 6.25
0.847 46.0 2254 0.8312 0.6767 0.4829 0.6175 0.6176 0.9136 0.9084 0.9105 8.125 16 4 12.2857 6.25
0.8588 47.0 2303 0.8306 0.6754 0.4814 0.6163 0.6163 0.9131 0.9082 0.9102 8.1473 16 4 12.3214 6.6964
0.8484 48.0 2352 0.8304 0.676 0.4816 0.6172 0.6179 0.9135 0.9082 0.9104 8.1161 16 4 12.2946 6.25
0.8514 49.0 2401 0.8303 0.676 0.4816 0.6172 0.6179 0.9135 0.9082 0.9104 8.1161 16 4 12.2946 6.25
0.8562 50.0 2450 0.8302 0.676 0.4816 0.6172 0.6179 0.9135 0.9082 0.9104 8.1161 16 4 12.2946 6.25

Framework versions

  • Transformers 4.33.1
  • Pytorch 2.0.1+cu118
  • Datasets 2.14.5
  • Tokenizers 0.13.3
Downloads last month
2
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_v57

Base model

google-t5/t5-small
Finetuned
(1502)
this model