hushem_5x_deit_tiny_adamax_00001_fold3
This model is a fine-tuned version of facebook/deit-tiny-patch16-224 on the imagefolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.5891
- Accuracy: 0.8605
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: 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 |
|---|---|---|---|---|
| 1.3491 | 1.0 | 28 | 1.3171 | 0.4186 |
| 1.0583 | 2.0 | 56 | 1.1404 | 0.4186 |
| 0.8133 | 3.0 | 84 | 1.0626 | 0.5581 |
| 0.7236 | 4.0 | 112 | 0.9689 | 0.6047 |
| 0.5407 | 5.0 | 140 | 0.9154 | 0.6512 |
| 0.4787 | 6.0 | 168 | 0.8329 | 0.6977 |
| 0.4043 | 7.0 | 196 | 0.7849 | 0.7442 |
| 0.3066 | 8.0 | 224 | 0.7047 | 0.7209 |
| 0.2483 | 9.0 | 252 | 0.6601 | 0.7209 |
| 0.1984 | 10.0 | 280 | 0.6346 | 0.7209 |
| 0.1508 | 11.0 | 308 | 0.6148 | 0.7209 |
| 0.1138 | 12.0 | 336 | 0.6034 | 0.7442 |
| 0.0962 | 13.0 | 364 | 0.5398 | 0.7674 |
| 0.0639 | 14.0 | 392 | 0.4866 | 0.7907 |
| 0.0434 | 15.0 | 420 | 0.4751 | 0.8140 |
| 0.0344 | 16.0 | 448 | 0.5249 | 0.7674 |
| 0.0259 | 17.0 | 476 | 0.4934 | 0.8140 |
| 0.0173 | 18.0 | 504 | 0.5157 | 0.8140 |
| 0.0125 | 19.0 | 532 | 0.4794 | 0.8140 |
| 0.0079 | 20.0 | 560 | 0.5000 | 0.8140 |
| 0.0068 | 21.0 | 588 | 0.5083 | 0.8140 |
| 0.0051 | 22.0 | 616 | 0.5005 | 0.8372 |
| 0.0044 | 23.0 | 644 | 0.4949 | 0.8372 |
| 0.0034 | 24.0 | 672 | 0.5221 | 0.8372 |
| 0.003 | 25.0 | 700 | 0.5304 | 0.8605 |
| 0.0025 | 26.0 | 728 | 0.5459 | 0.8372 |
| 0.0023 | 27.0 | 756 | 0.5309 | 0.8372 |
| 0.0022 | 28.0 | 784 | 0.5468 | 0.8605 |
| 0.002 | 29.0 | 812 | 0.5471 | 0.8372 |
| 0.0018 | 30.0 | 840 | 0.5437 | 0.8372 |
| 0.0015 | 31.0 | 868 | 0.5534 | 0.8372 |
| 0.0016 | 32.0 | 896 | 0.5689 | 0.8605 |
| 0.0015 | 33.0 | 924 | 0.5621 | 0.8605 |
| 0.0014 | 34.0 | 952 | 0.5754 | 0.8605 |
| 0.0013 | 35.0 | 980 | 0.5699 | 0.8605 |
| 0.0012 | 36.0 | 1008 | 0.5713 | 0.8605 |
| 0.0013 | 37.0 | 1036 | 0.5830 | 0.8372 |
| 0.0011 | 38.0 | 1064 | 0.5769 | 0.8372 |
| 0.0012 | 39.0 | 1092 | 0.5866 | 0.8372 |
| 0.0011 | 40.0 | 1120 | 0.5802 | 0.8372 |
| 0.0011 | 41.0 | 1148 | 0.5838 | 0.8605 |
| 0.001 | 42.0 | 1176 | 0.5874 | 0.8605 |
| 0.001 | 43.0 | 1204 | 0.5844 | 0.8605 |
| 0.001 | 44.0 | 1232 | 0.5856 | 0.8605 |
| 0.0009 | 45.0 | 1260 | 0.5886 | 0.8605 |
| 0.001 | 46.0 | 1288 | 0.5883 | 0.8605 |
| 0.0009 | 47.0 | 1316 | 0.5899 | 0.8605 |
| 0.0009 | 48.0 | 1344 | 0.5891 | 0.8605 |
| 0.001 | 49.0 | 1372 | 0.5891 | 0.8605 |
| 0.001 | 50.0 | 1400 | 0.5891 | 0.8605 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu118
- Datasets 2.15.0
- Tokenizers 0.15.0