deit_fold_3
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: 1.0647
- Accuracy: 0.9759
- F1 Score: 0.9781
- Recall: 0.9806
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: 1e-05
- train_batch_size: 64
- eval_batch_size: 64
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 150
- num_epochs: 100
- label_smoothing_factor: 0.15
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 Score | Recall |
|---|---|---|---|---|---|---|
| 2.8115 | 1.0 | 19 | 2.7735 | 0.3034 | 0.2529 | 0.2630 |
| 2.7100 | 2.0 | 38 | 2.6285 | 0.4069 | 0.2911 | 0.3284 |
| 2.5177 | 3.0 | 57 | 2.4014 | 0.5207 | 0.4125 | 0.4375 |
| 2.2202 | 4.0 | 76 | 2.0460 | 0.7069 | 0.7150 | 0.6830 |
| 1.7906 | 5.0 | 95 | 1.6558 | 0.8414 | 0.8552 | 0.8490 |
| 1.4703 | 6.0 | 114 | 1.4065 | 0.8724 | 0.8851 | 0.8883 |
| 1.2977 | 7.0 | 133 | 1.3043 | 0.9138 | 0.9200 | 0.9203 |
| 1.2124 | 8.0 | 152 | 1.2551 | 0.9345 | 0.9395 | 0.9427 |
| 1.1651 | 9.0 | 171 | 1.2009 | 0.9448 | 0.9490 | 0.9512 |
| 1.1434 | 10.0 | 190 | 1.1668 | 0.9483 | 0.9520 | 0.9536 |
| 1.1184 | 11.0 | 209 | 1.1541 | 0.9586 | 0.9611 | 0.9623 |
| 1.0736 | 12.0 | 228 | 1.1619 | 0.9517 | 0.9555 | 0.9598 |
| 1.0636 | 13.0 | 247 | 1.1813 | 0.9448 | 0.9499 | 0.9561 |
| 1.0464 | 14.0 | 266 | 1.1243 | 0.9655 | 0.9686 | 0.9721 |
| 1.0363 | 15.0 | 285 | 1.1079 | 0.9586 | 0.9615 | 0.9621 |
| 1.0317 | 16.0 | 304 | 1.1095 | 0.9655 | 0.9675 | 0.9708 |
| 1.0205 | 17.0 | 323 | 1.1498 | 0.9552 | 0.9593 | 0.9648 |
| 1.0146 | 18.0 | 342 | 1.1132 | 0.9655 | 0.9686 | 0.9721 |
| 1.0132 | 19.0 | 361 | 1.1042 | 0.9552 | 0.9586 | 0.9569 |
| 1.0060 | 20.0 | 380 | 1.0980 | 0.9552 | 0.9594 | 0.9549 |
| 0.9965 | 21.0 | 399 | 1.0731 | 0.9759 | 0.9780 | 0.9780 |
| 0.9954 | 22.0 | 418 | 1.0879 | 0.9621 | 0.9654 | 0.9631 |
| 1.0041 | 23.0 | 437 | 1.1282 | 0.9621 | 0.9656 | 0.9697 |
| 0.9945 | 24.0 | 456 | 1.0968 | 0.9655 | 0.9686 | 0.9721 |
| 0.9823 | 25.0 | 475 | 1.1093 | 0.9621 | 0.9652 | 0.9697 |
| 0.9811 | 26.0 | 494 | 1.0782 | 0.9690 | 0.9715 | 0.9746 |
| 0.9701 | 27.0 | 513 | 1.0829 | 0.9586 | 0.9615 | 0.9621 |
| 0.9695 | 28.0 | 532 | 1.0792 | 0.9724 | 0.9751 | 0.9782 |
| 0.9779 | 29.0 | 551 | 1.0625 | 0.9655 | 0.9684 | 0.9699 |
| 0.9825 | 30.0 | 570 | 1.0647 | 0.9759 | 0.9781 | 0.9806 |
| 0.9718 | 31.0 | 589 | 1.0718 | 0.9724 | 0.9750 | 0.9769 |
| 0.9798 | 32.0 | 608 | 1.0673 | 0.9759 | 0.9781 | 0.9806 |
| 0.9833 | 33.0 | 627 | 1.0644 | 0.9724 | 0.9750 | 0.9769 |
| 0.9665 | 34.0 | 646 | 1.0789 | 0.9724 | 0.9741 | 0.9770 |
| 0.9802 | 35.0 | 665 | 1.0593 | 0.9759 | 0.9781 | 0.9793 |
| 0.9624 | 36.0 | 684 | 1.0595 | 0.9759 | 0.9776 | 0.9806 |
| 0.9674 | 37.0 | 703 | 1.0814 | 0.9655 | 0.9681 | 0.9721 |
| 0.9723 | 38.0 | 722 | 1.0604 | 0.9759 | 0.9781 | 0.9806 |
| 0.9589 | 39.0 | 741 | 1.0768 | 0.9621 | 0.9655 | 0.9686 |
| 0.9655 | 40.0 | 760 | 1.0548 | 0.9655 | 0.9688 | 0.9694 |
| 0.9814 | 41.0 | 779 | 1.0676 | 0.9724 | 0.9751 | 0.9782 |
| 0.9627 | 42.0 | 798 | 1.0547 | 0.9759 | 0.9781 | 0.9806 |
| 0.9588 | 43.0 | 817 | 1.1152 | 0.9586 | 0.9626 | 0.9672 |
| 0.9665 | 44.0 | 836 | 1.0859 | 0.9690 | 0.9721 | 0.9757 |
| 0.9627 | 45.0 | 855 | 1.0683 | 0.9690 | 0.9721 | 0.9757 |
| 0.9576 | 46.0 | 874 | 1.0634 | 0.9759 | 0.9776 | 0.9806 |
| 0.9559 | 47.0 | 893 | 1.0620 | 0.9690 | 0.9718 | 0.9719 |
| 0.9642 | 48.0 | 912 | 1.0653 | 0.9724 | 0.9750 | 0.9756 |
| 0.9751 | 49.0 | 931 | 1.0574 | 0.9759 | 0.9781 | 0.9793 |
| 0.9688 | 50.0 | 950 | 1.0704 | 0.9690 | 0.9720 | 0.9744 |
| 0.9596 | 51.0 | 969 | 1.0635 | 0.9724 | 0.9750 | 0.9769 |
| 0.9716 | 52.0 | 988 | 1.0674 | 0.9759 | 0.9776 | 0.9806 |
Framework versions
- Transformers 5.0.0
- Pytorch 2.10.0+cu128
- Datasets 5.0.0
- Tokenizers 0.22.2