hushem_40x_deit_tiny_sgd_001_fold4
This model is a fine-tuned version of facebook/deit-tiny-patch16-224 on the imagefolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.2388
- Accuracy: 0.8810
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: 0.001
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- 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 |
|---|---|---|---|---|
| 1.2231 | 1.0 | 219 | 1.3846 | 0.2857 |
| 0.9485 | 2.0 | 438 | 1.1776 | 0.5238 |
| 0.8421 | 3.0 | 657 | 0.9985 | 0.6429 |
| 0.6802 | 4.0 | 876 | 0.8236 | 0.7381 |
| 0.5815 | 5.0 | 1095 | 0.6866 | 0.7857 |
| 0.5091 | 6.0 | 1314 | 0.5853 | 0.8095 |
| 0.3792 | 7.0 | 1533 | 0.5105 | 0.8333 |
| 0.3552 | 8.0 | 1752 | 0.4443 | 0.8333 |
| 0.3174 | 9.0 | 1971 | 0.4029 | 0.8810 |
| 0.2621 | 10.0 | 2190 | 0.3730 | 0.8571 |
| 0.2168 | 11.0 | 2409 | 0.3473 | 0.8571 |
| 0.2263 | 12.0 | 2628 | 0.3296 | 0.9048 |
| 0.1689 | 13.0 | 2847 | 0.3233 | 0.9048 |
| 0.171 | 14.0 | 3066 | 0.3040 | 0.8810 |
| 0.1176 | 15.0 | 3285 | 0.3059 | 0.8810 |
| 0.1241 | 16.0 | 3504 | 0.2811 | 0.8571 |
| 0.1343 | 17.0 | 3723 | 0.2712 | 0.8571 |
| 0.0953 | 18.0 | 3942 | 0.2802 | 0.8571 |
| 0.0918 | 19.0 | 4161 | 0.2700 | 0.8571 |
| 0.0691 | 20.0 | 4380 | 0.2755 | 0.8571 |
| 0.088 | 21.0 | 4599 | 0.2615 | 0.8571 |
| 0.0857 | 22.0 | 4818 | 0.2483 | 0.8571 |
| 0.0654 | 23.0 | 5037 | 0.2562 | 0.8571 |
| 0.0661 | 24.0 | 5256 | 0.2789 | 0.8571 |
| 0.0463 | 25.0 | 5475 | 0.2435 | 0.8571 |
| 0.0362 | 26.0 | 5694 | 0.2633 | 0.8571 |
| 0.0272 | 27.0 | 5913 | 0.2844 | 0.8571 |
| 0.041 | 28.0 | 6132 | 0.2942 | 0.8571 |
| 0.034 | 29.0 | 6351 | 0.2744 | 0.8571 |
| 0.0352 | 30.0 | 6570 | 0.2644 | 0.8810 |
| 0.0212 | 31.0 | 6789 | 0.2648 | 0.8810 |
| 0.0359 | 32.0 | 7008 | 0.2431 | 0.8810 |
| 0.0203 | 33.0 | 7227 | 0.2434 | 0.8810 |
| 0.0209 | 34.0 | 7446 | 0.2577 | 0.8810 |
| 0.0254 | 35.0 | 7665 | 0.2645 | 0.8810 |
| 0.0178 | 36.0 | 7884 | 0.2497 | 0.8810 |
| 0.0232 | 37.0 | 8103 | 0.2639 | 0.8810 |
| 0.015 | 38.0 | 8322 | 0.2391 | 0.8810 |
| 0.0246 | 39.0 | 8541 | 0.2615 | 0.8810 |
| 0.0228 | 40.0 | 8760 | 0.2445 | 0.8810 |
| 0.0203 | 41.0 | 8979 | 0.2448 | 0.8810 |
| 0.014 | 42.0 | 9198 | 0.2402 | 0.8810 |
| 0.0231 | 43.0 | 9417 | 0.2372 | 0.8810 |
| 0.0179 | 44.0 | 9636 | 0.2499 | 0.8810 |
| 0.0197 | 45.0 | 9855 | 0.2540 | 0.8810 |
| 0.0118 | 46.0 | 10074 | 0.2416 | 0.8810 |
| 0.0134 | 47.0 | 10293 | 0.2401 | 0.8810 |
| 0.0155 | 48.0 | 10512 | 0.2412 | 0.8810 |
| 0.0103 | 49.0 | 10731 | 0.2400 | 0.8810 |
| 0.0156 | 50.0 | 10950 | 0.2388 | 0.8810 |
Framework versions
- Transformers 4.32.1
- Pytorch 2.1.1+cu121
- Datasets 2.12.0
- Tokenizers 0.13.2