swin-tiny-patch4-window7-224-finetuned-swin-tiny
This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.5222
- Accuracy: 0.5559
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: 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: 50
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 3.5958 | 0.96 | 20 | 3.5209 | 0.0937 |
| 3.2466 | 1.98 | 41 | 2.9994 | 0.2387 |
| 2.4246 | 2.99 | 62 | 2.0341 | 0.4169 |
| 1.8599 | 4.0 | 83 | 1.6747 | 0.4955 |
| 1.531 | 4.96 | 103 | 1.5218 | 0.4773 |
| 1.3292 | 5.98 | 124 | 1.3834 | 0.5317 |
| 1.2063 | 6.99 | 145 | 1.3381 | 0.5468 |
| 1.0806 | 8.0 | 166 | 1.2748 | 0.5710 |
| 0.9638 | 8.96 | 186 | 1.3062 | 0.5559 |
| 0.8441 | 9.98 | 207 | 1.3322 | 0.5498 |
| 0.7868 | 10.99 | 228 | 1.2873 | 0.5710 |
| 0.7485 | 12.0 | 249 | 1.2012 | 0.5619 |
| 0.6522 | 12.96 | 269 | 1.2264 | 0.5861 |
| 0.6362 | 13.98 | 290 | 1.2796 | 0.5589 |
| 0.6214 | 14.99 | 311 | 1.3406 | 0.5529 |
| 0.5793 | 16.0 | 332 | 1.2479 | 0.5740 |
| 0.5187 | 16.96 | 352 | 1.3203 | 0.5891 |
| 0.4965 | 17.98 | 373 | 1.3429 | 0.5619 |
| 0.4809 | 18.99 | 394 | 1.3453 | 0.5831 |
| 0.4243 | 20.0 | 415 | 1.3759 | 0.5498 |
| 0.4447 | 20.96 | 435 | 1.4275 | 0.5196 |
| 0.3839 | 21.98 | 456 | 1.4660 | 0.5589 |
| 0.414 | 22.99 | 477 | 1.4465 | 0.5408 |
| 0.3741 | 24.0 | 498 | 1.3944 | 0.5650 |
| 0.3802 | 24.96 | 518 | 1.4272 | 0.5650 |
| 0.3733 | 25.98 | 539 | 1.3341 | 0.5589 |
| 0.3558 | 26.99 | 560 | 1.3864 | 0.5589 |
| 0.3448 | 28.0 | 581 | 1.4027 | 0.5589 |
| 0.3373 | 28.96 | 601 | 1.4452 | 0.5589 |
| 0.311 | 29.98 | 622 | 1.4021 | 0.5740 |
| 0.3218 | 30.99 | 643 | 1.4015 | 0.5680 |
| 0.3082 | 32.0 | 664 | 1.4159 | 0.5619 |
| 0.3173 | 32.96 | 684 | 1.4290 | 0.5498 |
| 0.2551 | 33.98 | 705 | 1.4268 | 0.5619 |
| 0.2739 | 34.99 | 726 | 1.4546 | 0.5559 |
| 0.2533 | 36.0 | 747 | 1.4398 | 0.5498 |
| 0.2578 | 36.96 | 767 | 1.4487 | 0.5438 |
| 0.2472 | 37.98 | 788 | 1.4438 | 0.5559 |
| 0.281 | 38.99 | 809 | 1.4916 | 0.5529 |
| 0.2757 | 40.0 | 830 | 1.4758 | 0.5619 |
| 0.2679 | 40.96 | 850 | 1.5104 | 0.5559 |
| 0.2548 | 41.98 | 871 | 1.5024 | 0.5529 |
| 0.2357 | 42.99 | 892 | 1.5286 | 0.5468 |
| 0.2357 | 44.0 | 913 | 1.5150 | 0.5529 |
| 0.2287 | 44.96 | 933 | 1.5234 | 0.5589 |
| 0.2329 | 45.98 | 954 | 1.5334 | 0.5650 |
| 0.2131 | 46.99 | 975 | 1.5296 | 0.5619 |
| 0.2269 | 48.0 | 996 | 1.5221 | 0.5559 |
| 0.2161 | 48.19 | 1000 | 1.5222 | 0.5559 |
Framework versions
- Transformers 4.37.0
- Pytorch 2.1.2
- Datasets 2.16.1
- Tokenizers 0.15.1