hushem_40x_deit_tiny_sgd_0001_fold3
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: 0.9959
- Accuracy: 0.6744
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
- 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 |
|---|---|---|---|---|
| 1.4187 | 1.0 | 217 | 1.4585 | 0.3023 |
| 1.3875 | 2.0 | 434 | 1.4331 | 0.2791 |
| 1.3134 | 3.0 | 651 | 1.4120 | 0.3256 |
| 1.3061 | 4.0 | 868 | 1.3930 | 0.3488 |
| 1.3015 | 5.0 | 1085 | 1.3762 | 0.3721 |
| 1.2507 | 6.0 | 1302 | 1.3597 | 0.3721 |
| 1.2542 | 7.0 | 1519 | 1.3427 | 0.3721 |
| 1.2153 | 8.0 | 1736 | 1.3276 | 0.3721 |
| 1.2187 | 9.0 | 1953 | 1.3127 | 0.4186 |
| 1.1894 | 10.0 | 2170 | 1.2981 | 0.4186 |
| 1.1545 | 11.0 | 2387 | 1.2843 | 0.4186 |
| 1.1296 | 12.0 | 2604 | 1.2697 | 0.4419 |
| 1.1425 | 13.0 | 2821 | 1.2546 | 0.4651 |
| 1.1006 | 14.0 | 3038 | 1.2420 | 0.4884 |
| 1.101 | 15.0 | 3255 | 1.2295 | 0.4884 |
| 1.0751 | 16.0 | 3472 | 1.2159 | 0.4884 |
| 1.0907 | 17.0 | 3689 | 1.2031 | 0.4884 |
| 1.047 | 18.0 | 3906 | 1.1903 | 0.5116 |
| 1.0396 | 19.0 | 4123 | 1.1781 | 0.5581 |
| 1.0151 | 20.0 | 4340 | 1.1663 | 0.5581 |
| 1.0071 | 21.0 | 4557 | 1.1547 | 0.5581 |
| 0.9605 | 22.0 | 4774 | 1.1441 | 0.5814 |
| 0.9825 | 23.0 | 4991 | 1.1328 | 0.6047 |
| 0.9877 | 24.0 | 5208 | 1.1238 | 0.6047 |
| 0.944 | 25.0 | 5425 | 1.1139 | 0.6047 |
| 1.0028 | 26.0 | 5642 | 1.1046 | 0.6047 |
| 0.9583 | 27.0 | 5859 | 1.0948 | 0.6279 |
| 0.9319 | 28.0 | 6076 | 1.0861 | 0.6279 |
| 0.8861 | 29.0 | 6293 | 1.0779 | 0.6279 |
| 0.9631 | 30.0 | 6510 | 1.0704 | 0.6512 |
| 0.8801 | 31.0 | 6727 | 1.0625 | 0.6512 |
| 0.9404 | 32.0 | 6944 | 1.0548 | 0.6512 |
| 0.9252 | 33.0 | 7161 | 1.0485 | 0.6512 |
| 0.8258 | 34.0 | 7378 | 1.0422 | 0.6512 |
| 0.8739 | 35.0 | 7595 | 1.0361 | 0.6744 |
| 0.8975 | 36.0 | 7812 | 1.0306 | 0.6744 |
| 0.8371 | 37.0 | 8029 | 1.0260 | 0.6744 |
| 0.8695 | 38.0 | 8246 | 1.0212 | 0.6744 |
| 0.8346 | 39.0 | 8463 | 1.0171 | 0.6744 |
| 0.8685 | 40.0 | 8680 | 1.0135 | 0.6744 |
| 0.8448 | 41.0 | 8897 | 1.0098 | 0.6744 |
| 0.8514 | 42.0 | 9114 | 1.0067 | 0.6744 |
| 0.8326 | 43.0 | 9331 | 1.0041 | 0.6744 |
| 0.8323 | 44.0 | 9548 | 1.0018 | 0.6744 |
| 0.8178 | 45.0 | 9765 | 0.9998 | 0.6744 |
| 0.8479 | 46.0 | 9982 | 0.9982 | 0.6744 |
| 0.8512 | 47.0 | 10199 | 0.9971 | 0.6744 |
| 0.851 | 48.0 | 10416 | 0.9963 | 0.6744 |
| 0.839 | 49.0 | 10633 | 0.9959 | 0.6744 |
| 0.7968 | 50.0 | 10850 | 0.9959 | 0.6744 |
Framework versions
- Transformers 4.32.1
- Pytorch 2.1.1+cu121
- Datasets 2.12.0
- Tokenizers 0.13.2