t5-small-finetuned-xsum-ashish-5000
This model is a fine-tuned version of t5-small on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 2.6200
- Rouge1: 14.8258
- Rouge2: 4.7741
- Rougel: 11.3583
- Rougelsum: 13.2147
- Gen Len: 19.0
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: 2e-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: 40
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|---|---|---|---|---|---|---|---|---|
| No log | 1.0 | 313 | 2.8460 | 13.7208 | 4.2759 | 10.4447 | 12.1604 | 19.0 |
| 2.8939 | 2.0 | 626 | 2.7686 | 14.0884 | 4.4571 | 10.8946 | 12.6399 | 19.0 |
| 2.8939 | 3.0 | 939 | 2.7323 | 14.249 | 4.4839 | 10.9701 | 12.7336 | 19.0 |
| 2.6857 | 4.0 | 1252 | 2.7140 | 14.4123 | 4.5447 | 11.09 | 12.8468 | 19.0 |
| 2.6353 | 5.0 | 1565 | 2.6962 | 14.4931 | 4.6524 | 11.1552 | 12.9235 | 19.0 |
| 2.6353 | 6.0 | 1878 | 2.6827 | 14.6765 | 4.6571 | 11.2099 | 13.0457 | 19.0 |
| 2.6005 | 7.0 | 2191 | 2.6743 | 14.6923 | 4.6506 | 11.1972 | 13.0305 | 19.0 |
| 2.5721 | 8.0 | 2504 | 2.6691 | 14.8242 | 4.7211 | 11.2794 | 13.1706 | 19.0 |
| 2.5721 | 9.0 | 2817 | 2.6598 | 14.9018 | 4.7961 | 11.3472 | 13.2632 | 19.0 |
| 2.5526 | 10.0 | 3130 | 2.6559 | 14.8855 | 4.8159 | 11.3402 | 13.2578 | 19.0 |
| 2.5526 | 11.0 | 3443 | 2.6533 | 14.8022 | 4.7367 | 11.2253 | 13.1308 | 19.0 |
| 2.5352 | 12.0 | 3756 | 2.6490 | 14.7306 | 4.6719 | 11.158 | 13.1083 | 19.0 |
| 2.5238 | 13.0 | 4069 | 2.6460 | 14.7908 | 4.6958 | 11.2061 | 13.1103 | 19.0 |
| 2.5238 | 14.0 | 4382 | 2.6436 | 14.7332 | 4.7132 | 11.1581 | 13.0709 | 19.0 |
| 2.5067 | 15.0 | 4695 | 2.6403 | 14.7062 | 4.7363 | 11.1275 | 13.0921 | 19.0 |
| 2.4922 | 16.0 | 5008 | 2.6382 | 14.735 | 4.6939 | 11.1301 | 13.0941 | 19.0 |
| 2.4922 | 17.0 | 5321 | 2.6353 | 14.8166 | 4.7615 | 11.2635 | 13.1526 | 19.0 |
| 2.4841 | 18.0 | 5634 | 2.6334 | 14.8517 | 4.8063 | 11.2705 | 13.1878 | 19.0 |
| 2.4841 | 19.0 | 5947 | 2.6306 | 14.7038 | 4.6747 | 11.1493 | 13.0818 | 19.0 |
| 2.4789 | 20.0 | 6260 | 2.6312 | 14.8127 | 4.7543 | 11.2775 | 13.1812 | 19.0 |
| 2.4644 | 21.0 | 6573 | 2.6285 | 14.7922 | 4.7114 | 11.2655 | 13.1716 | 19.0 |
| 2.4644 | 22.0 | 6886 | 2.6270 | 14.8587 | 4.78 | 11.3163 | 13.2017 | 19.0 |
| 2.4506 | 23.0 | 7199 | 2.6264 | 14.7304 | 4.6852 | 11.2138 | 13.1306 | 19.0 |
| 2.4595 | 24.0 | 7512 | 2.6258 | 14.7294 | 4.6597 | 11.2354 | 13.1126 | 19.0 |
| 2.4595 | 25.0 | 7825 | 2.6257 | 14.6318 | 4.6467 | 11.1913 | 13.0587 | 19.0 |
| 2.4523 | 26.0 | 8138 | 2.6250 | 14.7609 | 4.7037 | 11.2777 | 13.1711 | 19.0 |
| 2.4523 | 27.0 | 8451 | 2.6231 | 14.7342 | 4.7566 | 11.2569 | 13.1351 | 19.0 |
| 2.4317 | 28.0 | 8764 | 2.6223 | 14.725 | 4.7248 | 11.247 | 13.1234 | 19.0 |
| 2.4374 | 29.0 | 9077 | 2.6231 | 14.6911 | 4.7196 | 11.2372 | 13.0854 | 19.0 |
| 2.4374 | 30.0 | 9390 | 2.6234 | 14.6889 | 4.7202 | 11.2565 | 13.1003 | 19.0 |
| 2.4323 | 31.0 | 9703 | 2.6222 | 14.7264 | 4.7543 | 11.2752 | 13.1442 | 19.0 |
| 2.4295 | 32.0 | 10016 | 2.6215 | 14.7613 | 4.723 | 11.2632 | 13.1389 | 19.0 |
| 2.4295 | 33.0 | 10329 | 2.6212 | 14.7716 | 4.7676 | 11.3014 | 13.1637 | 19.0 |
| 2.4282 | 34.0 | 10642 | 2.6211 | 14.7547 | 4.7437 | 11.296 | 13.1552 | 19.0 |
| 2.4282 | 35.0 | 10955 | 2.6203 | 14.7717 | 4.7502 | 11.2999 | 13.1498 | 19.0 |
| 2.4265 | 36.0 | 11268 | 2.6208 | 14.7952 | 4.7795 | 11.3294 | 13.1866 | 19.0 |
| 2.4145 | 37.0 | 11581 | 2.6203 | 14.8122 | 4.7814 | 11.3385 | 13.1882 | 19.0 |
| 2.4145 | 38.0 | 11894 | 2.6202 | 14.8281 | 4.7798 | 11.3381 | 13.2065 | 19.0 |
| 2.4241 | 39.0 | 12207 | 2.6202 | 14.8163 | 4.7801 | 11.3492 | 13.2034 | 19.0 |
| 2.4163 | 40.0 | 12520 | 2.6200 | 14.8258 | 4.7741 | 11.3583 | 13.2147 | 19.0 |
Framework versions
- Transformers 4.17.0
- Pytorch 1.12.1+cu113
- Datasets 2.6.1
- Tokenizers 0.13.2