Adam_ViTL-16-224-1e-4-batch_16_epoch_4_classes_24
This model is a fine-tuned version of google/vit-large-patch16-224-in21k on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1561
- Accuracy: 0.9684
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: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 2
- mixed_precision_training: Native AMP
Training results
| Training Loss |
Epoch |
Step |
Validation Loss |
Accuracy |
| 0.7459 |
0.03 |
100 |
0.6501 |
0.8190 |
| 0.5929 |
0.07 |
200 |
0.4409 |
0.8836 |
| 0.2493 |
0.1 |
300 |
0.3525 |
0.9009 |
| 0.2142 |
0.14 |
400 |
0.3999 |
0.8779 |
| 0.3381 |
0.17 |
500 |
0.4229 |
0.8851 |
| 0.3445 |
0.21 |
600 |
0.2836 |
0.9195 |
| 0.2239 |
0.24 |
700 |
0.3989 |
0.8836 |
| 0.3475 |
0.28 |
800 |
0.2761 |
0.9210 |
| 0.0307 |
0.31 |
900 |
0.2963 |
0.9080 |
| 0.2957 |
0.35 |
1000 |
0.4865 |
0.8793 |
| 0.2431 |
0.38 |
1100 |
0.2740 |
0.9325 |
| 0.0729 |
0.42 |
1200 |
0.2630 |
0.9224 |
| 0.2757 |
0.45 |
1300 |
0.2515 |
0.9339 |
| 0.1763 |
0.49 |
1400 |
0.3826 |
0.9037 |
| 0.1481 |
0.52 |
1500 |
0.2282 |
0.9411 |
| 0.21 |
0.56 |
1600 |
0.2288 |
0.9454 |
| 0.2224 |
0.59 |
1700 |
0.3142 |
0.9296 |
| 0.0815 |
0.63 |
1800 |
0.2412 |
0.9411 |
| 0.0687 |
0.66 |
1900 |
0.2835 |
0.9353 |
| 0.3321 |
0.7 |
2000 |
0.3000 |
0.9282 |
| 0.1174 |
0.73 |
2100 |
0.2154 |
0.9440 |
| 0.0694 |
0.77 |
2200 |
0.2062 |
0.9497 |
| 0.0351 |
0.8 |
2300 |
0.1716 |
0.9511 |
| 0.088 |
0.84 |
2400 |
0.1410 |
0.9511 |
| 0.0856 |
0.87 |
2500 |
0.2342 |
0.9411 |
| 0.2248 |
0.91 |
2600 |
0.1954 |
0.9497 |
| 0.1188 |
0.94 |
2700 |
0.2655 |
0.9425 |
| 0.0322 |
0.98 |
2800 |
0.2535 |
0.9440 |
| 0.0739 |
1.01 |
2900 |
0.1640 |
0.9526 |
| 0.0352 |
1.04 |
3000 |
0.1760 |
0.9612 |
| 0.0007 |
1.08 |
3100 |
0.1593 |
0.9641 |
| 0.0107 |
1.11 |
3200 |
0.1970 |
0.9569 |
| 0.0027 |
1.15 |
3300 |
0.1603 |
0.9583 |
| 0.0005 |
1.18 |
3400 |
0.1550 |
0.9583 |
| 0.0637 |
1.22 |
3500 |
0.1874 |
0.9583 |
| 0.0006 |
1.25 |
3600 |
0.1829 |
0.9583 |
| 0.0626 |
1.29 |
3700 |
0.2311 |
0.9526 |
| 0.1023 |
1.32 |
3800 |
0.2325 |
0.9483 |
| 0.0014 |
1.36 |
3900 |
0.1556 |
0.9698 |
| 0.0186 |
1.39 |
4000 |
0.2151 |
0.9483 |
| 0.0005 |
1.43 |
4100 |
0.1369 |
0.9670 |
| 0.0005 |
1.46 |
4200 |
0.1240 |
0.9727 |
| 0.0004 |
1.5 |
4300 |
0.2019 |
0.9612 |
| 0.0008 |
1.53 |
4400 |
0.1361 |
0.9713 |
| 0.013 |
1.57 |
4500 |
0.1343 |
0.9684 |
| 0.014 |
1.6 |
4600 |
0.1553 |
0.9670 |
| 0.0005 |
1.64 |
4700 |
0.1528 |
0.9655 |
| 0.0003 |
1.67 |
4800 |
0.1586 |
0.9641 |
| 0.0009 |
1.71 |
4900 |
0.1598 |
0.9655 |
| 0.0003 |
1.74 |
5000 |
0.1727 |
0.9641 |
| 0.0003 |
1.78 |
5100 |
0.1521 |
0.9727 |
| 0.0076 |
1.81 |
5200 |
0.1534 |
0.9698 |
| 0.0003 |
1.85 |
5300 |
0.1656 |
0.9655 |
| 0.0003 |
1.88 |
5400 |
0.1833 |
0.9641 |
| 0.0003 |
1.92 |
5500 |
0.1719 |
0.9670 |
| 0.0003 |
1.95 |
5600 |
0.1565 |
0.9684 |
| 0.0003 |
1.99 |
5700 |
0.1561 |
0.9684 |
Framework versions
- Transformers 4.39.3
- Pytorch 2.1.2
- Datasets 2.18.0
- Tokenizers 0.15.2
