deit_fold_1_v3
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: 1.1628
- Accuracy: 0.9551
- F1 Score: 0.9570
- Recall: 0.9581
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 |
|---|---|---|---|---|---|---|
| 2.8330 | 1.0 | 20 | 2.7860 | 0.2372 | 0.2351 | 0.2394 |
| 2.6641 | 2.0 | 40 | 2.6204 | 0.3942 | 0.3682 | 0.3639 |
| 2.4039 | 3.0 | 60 | 2.3714 | 0.5705 | 0.5479 | 0.5415 |
| 2.0152 | 4.0 | 80 | 2.0046 | 0.7308 | 0.7211 | 0.7286 |
| 1.6639 | 5.0 | 100 | 1.6679 | 0.8013 | 0.8035 | 0.8072 |
| 1.4446 | 6.0 | 120 | 1.5218 | 0.8301 | 0.8312 | 0.8445 |
| 1.3294 | 7.0 | 140 | 1.4034 | 0.8814 | 0.8808 | 0.8851 |
| 1.2539 | 8.0 | 160 | 1.3367 | 0.9071 | 0.9055 | 0.9059 |
| 1.1603 | 9.0 | 180 | 1.2977 | 0.9167 | 0.9155 | 0.9151 |
| 1.1220 | 10.0 | 200 | 1.2600 | 0.9167 | 0.9148 | 0.9169 |
| 1.1301 | 11.0 | 220 | 1.2411 | 0.9263 | 0.9260 | 0.9256 |
| 1.0514 | 12.0 | 240 | 1.2276 | 0.9167 | 0.9162 | 0.9200 |
| 1.0528 | 13.0 | 260 | 1.2156 | 0.9359 | 0.9364 | 0.9360 |
| 1.0386 | 14.0 | 280 | 1.2090 | 0.9327 | 0.9324 | 0.9317 |
| 1.0265 | 15.0 | 300 | 1.1935 | 0.9295 | 0.9303 | 0.9317 |
| 1.0239 | 16.0 | 320 | 1.1888 | 0.9391 | 0.9397 | 0.9392 |
| 1.0195 | 17.0 | 340 | 1.2041 | 0.9295 | 0.9304 | 0.9311 |
| 1.0003 | 18.0 | 360 | 1.1992 | 0.9423 | 0.9425 | 0.9409 |
| 1.0130 | 19.0 | 380 | 1.1703 | 0.9455 | 0.9478 | 0.9483 |
| 1.0041 | 20.0 | 400 | 1.1874 | 0.9391 | 0.9393 | 0.9384 |
| 1.0013 | 21.0 | 420 | 1.1704 | 0.9359 | 0.9367 | 0.9360 |
| 0.9889 | 22.0 | 440 | 1.1805 | 0.9423 | 0.9447 | 0.9483 |
| 0.9880 | 23.0 | 460 | 1.1663 | 0.9423 | 0.9439 | 0.9427 |
| 0.9856 | 24.0 | 480 | 1.1758 | 0.9455 | 0.9474 | 0.9458 |
| 0.9739 | 25.0 | 500 | 1.1704 | 0.9455 | 0.9474 | 0.9458 |
| 0.9741 | 26.0 | 520 | 1.1529 | 0.9519 | 0.9542 | 0.9563 |
| 0.9671 | 27.0 | 540 | 1.1938 | 0.9455 | 0.9472 | 0.9449 |
| 0.9772 | 28.0 | 560 | 1.1926 | 0.9423 | 0.9444 | 0.9434 |
| 0.9697 | 29.0 | 580 | 1.1817 | 0.9359 | 0.9374 | 0.9362 |
| 0.9611 | 30.0 | 600 | 1.1628 | 0.9551 | 0.9570 | 0.9581 |
| 0.9648 | 31.0 | 620 | 1.1926 | 0.9455 | 0.9476 | 0.9471 |
| 0.9619 | 32.0 | 640 | 1.1731 | 0.9519 | 0.9535 | 0.9550 |
| 0.9647 | 33.0 | 660 | 1.1810 | 0.9519 | 0.9537 | 0.9562 |
| 0.9632 | 34.0 | 680 | 1.1731 | 0.9551 | 0.9566 | 0.9586 |
| 0.9644 | 35.0 | 700 | 1.1879 | 0.9455 | 0.9474 | 0.9471 |
| 0.9615 | 36.0 | 720 | 1.1760 | 0.9519 | 0.9537 | 0.9562 |
Framework versions
- Transformers 5.0.0
- Pytorch 2.10.0+cu128
- Datasets 5.0.0
- Tokenizers 0.22.2