swin-tiny-patch4-window7-224-finetuned-eurosat
This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 on the imagefolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.1073
- Accuracy: 0.9675
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: 5e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 30
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| No log | 1.0 | 5 | 0.6616 | 0.6299 |
| 0.6583 | 2.0 | 10 | 0.5232 | 0.7597 |
| 0.6583 | 3.0 | 15 | 0.5043 | 0.7857 |
| 0.3346 | 4.0 | 20 | 0.2879 | 0.8766 |
| 0.3346 | 5.0 | 25 | 0.2424 | 0.9091 |
| 0.1544 | 6.0 | 30 | 0.2217 | 0.8896 |
| 0.1544 | 7.0 | 35 | 0.1466 | 0.9221 |
| 0.088 | 8.0 | 40 | 0.1261 | 0.9481 |
| 0.088 | 9.0 | 45 | 0.1680 | 0.9221 |
| 0.0977 | 10.0 | 50 | 0.1446 | 0.9351 |
| 0.0977 | 11.0 | 55 | 0.1812 | 0.9221 |
| 0.0719 | 12.0 | 60 | 0.1798 | 0.9286 |
| 0.0719 | 13.0 | 65 | 0.1056 | 0.9610 |
| 0.0629 | 14.0 | 70 | 0.1073 | 0.9675 |
| 0.0629 | 15.0 | 75 | 0.1106 | 0.9545 |
| 0.0414 | 16.0 | 80 | 0.1286 | 0.9416 |
| 0.0414 | 17.0 | 85 | 0.0761 | 0.9610 |
| 0.0397 | 18.0 | 90 | 0.0785 | 0.9675 |
| 0.0397 | 19.0 | 95 | 0.0746 | 0.9675 |
| 0.0487 | 20.0 | 100 | 0.0684 | 0.9675 |
| 0.0487 | 21.0 | 105 | 0.0602 | 0.9610 |
| 0.0244 | 22.0 | 110 | 0.0551 | 0.9675 |
| 0.0244 | 23.0 | 115 | 0.0639 | 0.9675 |
| 0.0214 | 24.0 | 120 | 0.0583 | 0.9675 |
| 0.0214 | 25.0 | 125 | 0.0663 | 0.9675 |
| 0.0261 | 26.0 | 130 | 0.1006 | 0.9610 |
| 0.0261 | 27.0 | 135 | 0.0711 | 0.9675 |
| 0.019 | 28.0 | 140 | 0.0629 | 0.9675 |
| 0.019 | 29.0 | 145 | 0.0728 | 0.9610 |
| 0.0237 | 30.0 | 150 | 0.0747 | 0.9610 |
Framework versions
- Transformers 4.38.1
- Pytorch 1.10.0+cu111
- Datasets 2.17.1
- Tokenizers 0.15.2