swin-tiny-patch4-window7-224-finetuned-eurosat
This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.1792
- Accuracy: 0.4627
- F1: 0.1004
- Precision: 0.2132
- Recall: 0.0852
- Auc Roc: 0.7403
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: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 32
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 40
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 | Precision | Recall | Auc Roc |
|---|---|---|---|---|---|---|---|---|
| 0.2828 | 1.0 | 82 | 0.2357 | 0.3645 | 0.0412 | 0.0410 | 0.0414 | 0.5445 |
| 0.2048 | 2.0 | 164 | 0.1922 | 0.5713 | 0.0492 | 0.0396 | 0.0648 | 0.5762 |
| 0.2067 | 3.0 | 246 | 0.1899 | 0.4824 | 0.0493 | 0.1107 | 0.0550 | 0.6240 |
| 0.1974 | 4.0 | 328 | 0.1870 | 0.4079 | 0.0474 | 0.0485 | 0.0463 | 0.6619 |
| 0.2021 | 5.0 | 410 | 0.2048 | 0.1612 | 0.0358 | 0.1222 | 0.0227 | 0.6841 |
| 0.1843 | 6.0 | 492 | 0.1838 | 0.3709 | 0.0625 | 0.0967 | 0.0526 | 0.6965 |
| 0.187 | 7.0 | 574 | 0.1800 | 0.4009 | 0.0610 | 0.0963 | 0.0535 | 0.7127 |
| 0.1992 | 8.0 | 656 | 0.1775 | 0.4547 | 0.0563 | 0.1095 | 0.0548 | 0.7172 |
| 0.179 | 9.0 | 738 | 0.1863 | 0.3316 | 0.0667 | 0.1044 | 0.0521 | 0.7192 |
| 0.1683 | 10.0 | 820 | 0.1786 | 0.4165 | 0.0573 | 0.0951 | 0.0521 | 0.7194 |
| 0.1742 | 11.0 | 902 | 0.1764 | 0.4575 | 0.0714 | 0.1289 | 0.0640 | 0.7240 |
| 0.1871 | 12.0 | 984 | 0.1767 | 0.5130 | 0.0626 | 0.1030 | 0.0642 | 0.7272 |
| 0.1802 | 13.0 | 1066 | 0.1750 | 0.4575 | 0.0604 | 0.1088 | 0.0573 | 0.7305 |
| 0.1666 | 14.0 | 1148 | 0.1755 | 0.4414 | 0.0847 | 0.1338 | 0.0732 | 0.7282 |
| 0.1674 | 15.0 | 1230 | 0.1767 | 0.4263 | 0.0733 | 0.0990 | 0.0651 | 0.7296 |
| 0.1846 | 16.0 | 1312 | 0.1771 | 0.4460 | 0.0866 | 0.1363 | 0.0784 | 0.7349 |
| 0.1819 | 17.0 | 1394 | 0.1762 | 0.4356 | 0.0686 | 0.1063 | 0.0604 | 0.7282 |
| 0.1594 | 18.0 | 1476 | 0.1757 | 0.4298 | 0.0804 | 0.1289 | 0.0699 | 0.7345 |
| 0.1805 | 19.0 | 1558 | 0.1783 | 0.4206 | 0.0884 | 0.1358 | 0.0803 | 0.7323 |
| 0.164 | 20.0 | 1640 | 0.1768 | 0.4858 | 0.0896 | 0.1683 | 0.0816 | 0.7377 |
| 0.1673 | 21.0 | 1722 | 0.1779 | 0.4292 | 0.0888 | 0.1500 | 0.0764 | 0.7281 |
| 0.1719 | 22.0 | 1804 | 0.1780 | 0.5142 | 0.0783 | 0.1409 | 0.0746 | 0.7358 |
| 0.1598 | 23.0 | 1886 | 0.1762 | 0.4801 | 0.0876 | 0.2268 | 0.0778 | 0.7387 |
| 0.1682 | 24.0 | 1968 | 0.1788 | 0.4570 | 0.0954 | 0.1420 | 0.0849 | 0.7406 |
| 0.1558 | 25.0 | 2050 | 0.1795 | 0.4391 | 0.1023 | 0.2186 | 0.0912 | 0.7396 |
| 0.1695 | 26.0 | 2132 | 0.1768 | 0.4830 | 0.0881 | 0.2050 | 0.0789 | 0.7368 |
| 0.1514 | 27.0 | 2214 | 0.1771 | 0.4402 | 0.0958 | 0.2160 | 0.0807 | 0.7385 |
| 0.1669 | 28.0 | 2296 | 0.1773 | 0.4876 | 0.0892 | 0.2195 | 0.0791 | 0.7428 |
| 0.156 | 29.0 | 2378 | 0.1802 | 0.4107 | 0.1000 | 0.2148 | 0.0820 | 0.7331 |
| 0.1555 | 30.0 | 2460 | 0.1795 | 0.4258 | 0.0975 | 0.2033 | 0.0832 | 0.7336 |
| 0.1527 | 31.0 | 2542 | 0.1773 | 0.4853 | 0.0955 | 0.2149 | 0.0835 | 0.7407 |
| 0.1592 | 32.0 | 2624 | 0.1786 | 0.4830 | 0.0920 | 0.2138 | 0.0810 | 0.7402 |
| 0.1567 | 33.0 | 2706 | 0.1788 | 0.4916 | 0.0908 | 0.2125 | 0.0795 | 0.7415 |
| 0.1595 | 34.0 | 2788 | 0.1788 | 0.4558 | 0.0993 | 0.2175 | 0.0835 | 0.7385 |
| 0.1431 | 35.0 | 2870 | 0.1791 | 0.4529 | 0.1043 | 0.2087 | 0.0867 | 0.7364 |
| 0.1709 | 36.0 | 2952 | 0.1792 | 0.4541 | 0.1009 | 0.2145 | 0.0861 | 0.7389 |
| 0.1525 | 37.0 | 3034 | 0.1789 | 0.4714 | 0.0957 | 0.2123 | 0.0832 | 0.7405 |
| 0.1467 | 38.0 | 3116 | 0.1791 | 0.4639 | 0.0981 | 0.2110 | 0.0840 | 0.7397 |
| 0.1531 | 39.0 | 3198 | 0.1791 | 0.4633 | 0.1000 | 0.2123 | 0.0850 | 0.7404 |
| 0.1507 | 39.5169 | 3240 | 0.1792 | 0.4627 | 0.1004 | 0.2132 | 0.0852 | 0.7403 |
Framework versions
- Transformers 4.51.3
- Pytorch 2.6.0+cu124
- Datasets 3.5.0
- Tokenizers 0.21.1