swin-tiny-patch4-window7-224-celeba-smiling
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: 0.1561
- Accuracy: 0.9357
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: 128
- eval_batch_size: 64
- seed: 42
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 0.1
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 0.2037 | 0.3931 | 500 | 0.1882 | 0.9233 |
| 0.1872 | 0.7862 | 1000 | 0.1701 | 0.9291 |
| 0.1802 | 1.1792 | 1500 | 0.1846 | 0.9232 |
| 0.1714 | 1.5723 | 2000 | 0.1665 | 0.9310 |
| 0.1614 | 1.9654 | 2500 | 0.1603 | 0.9311 |
| 0.1590 | 2.3585 | 3000 | 0.1557 | 0.9341 |
| 0.1532 | 2.7516 | 3500 | 0.1561 | 0.9357 |
| 0.1443 | 3.1447 | 4000 | 0.1581 | 0.9328 |
| 0.1482 | 3.5377 | 4500 | 0.1685 | 0.9315 |
| 0.1496 | 3.9308 | 5000 | 0.1537 | 0.9352 |
| 0.1362 | 4.3239 | 5500 | 0.1568 | 0.9357 |
| 0.1338 | 4.7170 | 6000 | 0.1719 | 0.9324 |
| 0.1200 | 5.1101 | 6500 | 0.1848 | 0.9267 |
| 0.1197 | 5.5031 | 7000 | 0.1699 | 0.9315 |
| 0.1194 | 5.8962 | 7500 | 0.1665 | 0.9324 |
| 0.1026 | 6.2893 | 8000 | 0.1867 | 0.9293 |
| 0.1007 | 6.6824 | 8500 | 0.1842 | 0.9315 |
| 0.0867 | 7.0755 | 9000 | 0.2063 | 0.9297 |
| 0.0809 | 7.4686 | 9500 | 0.2178 | 0.9307 |
| 0.0843 | 7.8616 | 10000 | 0.2200 | 0.9272 |
| 0.0778 | 8.2547 | 10500 | 0.2253 | 0.9296 |
| 0.0738 | 8.6478 | 11000 | 0.2282 | 0.9295 |
| 0.0723 | 9.0409 | 11500 | 0.2353 | 0.9306 |
| 0.0697 | 9.4340 | 12000 | 0.2333 | 0.9300 |
| 0.0661 | 9.8270 | 12500 | 0.2336 | 0.9302 |
Framework versions
- Transformers 5.3.0
- Pytorch 2.6.0+cu124
- Datasets 4.8.3
- Tokenizers 0.22.2