hushem_40x_deit_small_sgd_001_fold4
This model is a fine-tuned version of facebook/deit-small-patch16-224 on the imagefolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.2365
- 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 |
|---|---|---|---|---|
| 1.2757 | 1.0 | 219 | 1.3298 | 0.2619 |
| 1.0766 | 2.0 | 438 | 1.1919 | 0.4048 |
| 0.9095 | 3.0 | 657 | 1.0786 | 0.5476 |
| 0.7507 | 4.0 | 876 | 0.9821 | 0.5476 |
| 0.6994 | 5.0 | 1095 | 0.8850 | 0.5952 |
| 0.5864 | 6.0 | 1314 | 0.8204 | 0.6429 |
| 0.4328 | 7.0 | 1533 | 0.7576 | 0.6905 |
| 0.4293 | 8.0 | 1752 | 0.6999 | 0.7143 |
| 0.3464 | 9.0 | 1971 | 0.6320 | 0.7143 |
| 0.3175 | 10.0 | 2190 | 0.5956 | 0.7381 |
| 0.2382 | 11.0 | 2409 | 0.5588 | 0.7381 |
| 0.2672 | 12.0 | 2628 | 0.5195 | 0.7381 |
| 0.2016 | 13.0 | 2847 | 0.4850 | 0.8095 |
| 0.1832 | 14.0 | 3066 | 0.4528 | 0.8095 |
| 0.1406 | 15.0 | 3285 | 0.4338 | 0.8333 |
| 0.1305 | 16.0 | 3504 | 0.3948 | 0.8571 |
| 0.1504 | 17.0 | 3723 | 0.3785 | 0.8571 |
| 0.1139 | 18.0 | 3942 | 0.3689 | 0.8571 |
| 0.096 | 19.0 | 4161 | 0.3548 | 0.8571 |
| 0.0869 | 20.0 | 4380 | 0.3393 | 0.8571 |
| 0.0874 | 21.0 | 4599 | 0.3057 | 0.8571 |
| 0.0797 | 22.0 | 4818 | 0.2990 | 0.8571 |
| 0.0596 | 23.0 | 5037 | 0.2862 | 0.8571 |
| 0.053 | 24.0 | 5256 | 0.3012 | 0.8810 |
| 0.0562 | 25.0 | 5475 | 0.2885 | 0.8810 |
| 0.0463 | 26.0 | 5694 | 0.2676 | 0.8810 |
| 0.0374 | 27.0 | 5913 | 0.2870 | 0.8810 |
| 0.037 | 28.0 | 6132 | 0.2638 | 0.8810 |
| 0.0341 | 29.0 | 6351 | 0.2690 | 0.8810 |
| 0.0327 | 30.0 | 6570 | 0.2566 | 0.8810 |
| 0.0238 | 31.0 | 6789 | 0.2611 | 0.8810 |
| 0.0256 | 32.0 | 7008 | 0.2643 | 0.8810 |
| 0.0284 | 33.0 | 7227 | 0.2717 | 0.8810 |
| 0.0213 | 34.0 | 7446 | 0.2627 | 0.8810 |
| 0.0191 | 35.0 | 7665 | 0.2395 | 0.8810 |
| 0.0246 | 36.0 | 7884 | 0.2517 | 0.8810 |
| 0.0207 | 37.0 | 8103 | 0.2515 | 0.8810 |
| 0.0134 | 38.0 | 8322 | 0.2484 | 0.8810 |
| 0.0162 | 39.0 | 8541 | 0.2279 | 0.8810 |
| 0.0165 | 40.0 | 8760 | 0.2516 | 0.8810 |
| 0.0146 | 41.0 | 8979 | 0.2253 | 0.8810 |
| 0.0168 | 42.0 | 9198 | 0.2425 | 0.8810 |
| 0.0155 | 43.0 | 9417 | 0.2370 | 0.8810 |
| 0.0145 | 44.0 | 9636 | 0.2352 | 0.8810 |
| 0.0118 | 45.0 | 9855 | 0.2414 | 0.8810 |
| 0.0107 | 46.0 | 10074 | 0.2338 | 0.8810 |
| 0.0124 | 47.0 | 10293 | 0.2350 | 0.8810 |
| 0.0125 | 48.0 | 10512 | 0.2352 | 0.8810 |
| 0.0138 | 49.0 | 10731 | 0.2367 | 0.8810 |
| 0.0183 | 50.0 | 10950 | 0.2365 | 0.8810 |
Framework versions
- Transformers 4.32.1
- Pytorch 2.1.0+cu121
- Datasets 2.12.0
- Tokenizers 0.13.2