vit_fold_2_v3
This model is a fine-tuned version of google/vit-base-patch16-224 on the imagefolder dataset. It achieves the following results on the evaluation set:
- Loss: 1.1369
- Accuracy: 0.9551
- F1 Score: 0.9563
- Recall: 0.9560
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 |
|---|---|---|---|---|---|---|
| 2.8711 | 1.0 | 20 | 2.9097 | 0.2532 | 0.2497 | 0.2611 |
| 2.6539 | 2.0 | 40 | 2.6799 | 0.3910 | 0.3842 | 0.3928 |
| 2.2936 | 3.0 | 60 | 2.3974 | 0.5609 | 0.5597 | 0.5774 |
| 1.8549 | 4.0 | 80 | 2.0469 | 0.7115 | 0.7136 | 0.7281 |
| 1.5368 | 5.0 | 100 | 1.6885 | 0.8141 | 0.8169 | 0.8230 |
| 1.3205 | 6.0 | 120 | 1.4580 | 0.8814 | 0.8823 | 0.8864 |
| 1.1521 | 7.0 | 140 | 1.3361 | 0.9006 | 0.9020 | 0.9004 |
| 1.1270 | 8.0 | 160 | 1.2874 | 0.9103 | 0.9118 | 0.9105 |
| 1.0616 | 9.0 | 180 | 1.2559 | 0.9231 | 0.9242 | 0.9269 |
| 1.0597 | 10.0 | 200 | 1.2492 | 0.9071 | 0.9075 | 0.9049 |
| 1.0621 | 11.0 | 220 | 1.2381 | 0.9103 | 0.9096 | 0.9073 |
| 1.0121 | 12.0 | 240 | 1.2098 | 0.9231 | 0.9238 | 0.9224 |
| 1.0036 | 13.0 | 260 | 1.1897 | 0.9391 | 0.9392 | 0.9408 |
| 1.0230 | 14.0 | 280 | 1.1786 | 0.9327 | 0.9335 | 0.9318 |
| 0.9965 | 15.0 | 300 | 1.1717 | 0.9455 | 0.9465 | 0.9481 |
| 1.0003 | 16.0 | 320 | 1.1839 | 0.9327 | 0.9323 | 0.9313 |
| 0.9968 | 17.0 | 340 | 1.1635 | 0.9423 | 0.9438 | 0.9434 |
| 0.9940 | 18.0 | 360 | 1.1561 | 0.9359 | 0.9375 | 0.9395 |
| 0.9948 | 19.0 | 380 | 1.1597 | 0.9423 | 0.9438 | 0.9434 |
| 0.9795 | 20.0 | 400 | 1.1576 | 0.9455 | 0.9467 | 0.9459 |
| 0.9752 | 21.0 | 420 | 1.1580 | 0.9391 | 0.9403 | 0.9417 |
| 0.9768 | 22.0 | 440 | 1.1596 | 0.9423 | 0.9435 | 0.9422 |
| 0.9644 | 23.0 | 460 | 1.1587 | 0.9359 | 0.9372 | 0.9370 |
| 0.9746 | 24.0 | 480 | 1.1708 | 0.9455 | 0.9460 | 0.9434 |
| 0.9711 | 25.0 | 500 | 1.1564 | 0.9391 | 0.9406 | 0.9398 |
| 0.9689 | 26.0 | 520 | 1.1522 | 0.9391 | 0.9406 | 0.9398 |
| 0.9710 | 27.0 | 540 | 1.1525 | 0.9455 | 0.9466 | 0.9447 |
| 0.9754 | 28.0 | 560 | 1.1468 | 0.9455 | 0.9467 | 0.9459 |
| 0.9686 | 29.0 | 580 | 1.1488 | 0.9455 | 0.9466 | 0.9451 |
| 0.9678 | 30.0 | 600 | 1.1476 | 0.9487 | 0.9496 | 0.9483 |
| 0.9690 | 31.0 | 620 | 1.1462 | 0.9455 | 0.9467 | 0.9459 |
| 0.9630 | 32.0 | 640 | 1.1533 | 0.9487 | 0.9494 | 0.9471 |
| 0.9621 | 33.0 | 660 | 1.1516 | 0.9551 | 0.9555 | 0.9532 |
| 0.9670 | 34.0 | 680 | 1.1439 | 0.9487 | 0.9502 | 0.9499 |
| 0.9765 | 35.0 | 700 | 1.1464 | 0.9487 | 0.9502 | 0.9499 |
| 0.9769 | 36.0 | 720 | 1.1411 | 0.9455 | 0.9467 | 0.9459 |
| 0.9649 | 37.0 | 740 | 1.1353 | 0.9487 | 0.9496 | 0.9483 |
| 0.9563 | 38.0 | 760 | 1.1369 | 0.9551 | 0.9563 | 0.9560 |
| 0.9635 | 39.0 | 780 | 1.1267 | 0.9519 | 0.9533 | 0.9535 |
| 0.9667 | 40.0 | 800 | 1.1389 | 0.9455 | 0.9464 | 0.9447 |
| 0.9635 | 41.0 | 820 | 1.1282 | 0.9487 | 0.9497 | 0.9496 |
| 0.9552 | 42.0 | 840 | 1.1330 | 0.9551 | 0.9555 | 0.9532 |
| 0.9688 | 43.0 | 860 | 1.1279 | 0.9519 | 0.9525 | 0.9508 |
| 0.9655 | 44.0 | 880 | 1.1347 | 0.9551 | 0.9563 | 0.9560 |
| 0.9648 | 45.0 | 900 | 1.1343 | 0.9519 | 0.9531 | 0.9523 |
| 0.9791 | 46.0 | 920 | 1.1304 | 0.9551 | 0.9557 | 0.9545 |
| 0.9625 | 47.0 | 940 | 1.1320 | 0.9487 | 0.9499 | 0.9508 |
| 0.9636 | 48.0 | 960 | 1.1402 | 0.9519 | 0.9525 | 0.9508 |
| 0.9667 | 49.0 | 980 | 1.1401 | 0.9551 | 0.9561 | 0.9548 |
Framework versions
- Transformers 5.0.0
- Pytorch 2.10.0+cu128
- Datasets 5.0.0
- Tokenizers 0.22.2