hushem_40x_deit_small_rms_001_fold5
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: 1.6080
- Accuracy: 0.7805
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: 0.001
- 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 |
|---|---|---|---|---|
| 1.4202 | 1.0 | 220 | 1.3858 | 0.2439 |
| 1.4846 | 2.0 | 440 | 1.3050 | 0.3659 |
| 1.213 | 3.0 | 660 | 1.2527 | 0.3659 |
| 1.2436 | 4.0 | 880 | 1.2166 | 0.3659 |
| 1.1042 | 5.0 | 1100 | 1.0859 | 0.4878 |
| 1.1247 | 6.0 | 1320 | 1.0109 | 0.5854 |
| 1.1005 | 7.0 | 1540 | 1.0680 | 0.4634 |
| 1.0292 | 8.0 | 1760 | 0.9693 | 0.5366 |
| 0.9154 | 9.0 | 1980 | 1.0620 | 0.6829 |
| 0.8944 | 10.0 | 2200 | 0.7963 | 0.6585 |
| 0.9136 | 11.0 | 2420 | 1.0510 | 0.5854 |
| 0.8882 | 12.0 | 2640 | 0.9455 | 0.6098 |
| 0.8527 | 13.0 | 2860 | 0.7251 | 0.6098 |
| 0.7429 | 14.0 | 3080 | 0.7653 | 0.7805 |
| 0.6936 | 15.0 | 3300 | 0.7806 | 0.7561 |
| 0.7049 | 16.0 | 3520 | 1.3071 | 0.6585 |
| 0.6294 | 17.0 | 3740 | 0.5950 | 0.7561 |
| 0.6472 | 18.0 | 3960 | 0.4686 | 0.7561 |
| 0.6452 | 19.0 | 4180 | 0.5722 | 0.7561 |
| 0.5301 | 20.0 | 4400 | 0.8370 | 0.6341 |
| 0.5967 | 21.0 | 4620 | 0.5708 | 0.8537 |
| 0.4666 | 22.0 | 4840 | 0.3950 | 0.8293 |
| 0.5239 | 23.0 | 5060 | 0.5273 | 0.8049 |
| 0.5888 | 24.0 | 5280 | 0.4686 | 0.7561 |
| 0.5034 | 25.0 | 5500 | 0.5275 | 0.7561 |
| 0.5755 | 26.0 | 5720 | 0.4866 | 0.7805 |
| 0.5913 | 27.0 | 5940 | 0.8503 | 0.7805 |
| 0.405 | 28.0 | 6160 | 0.5132 | 0.8049 |
| 0.4508 | 29.0 | 6380 | 0.7683 | 0.7805 |
| 0.4334 | 30.0 | 6600 | 0.7092 | 0.8293 |
| 0.4219 | 31.0 | 6820 | 0.5130 | 0.7561 |
| 0.4182 | 32.0 | 7040 | 0.5153 | 0.8293 |
| 0.308 | 33.0 | 7260 | 0.5501 | 0.7561 |
| 0.3445 | 34.0 | 7480 | 0.6037 | 0.8049 |
| 0.2565 | 35.0 | 7700 | 0.5862 | 0.7805 |
| 0.2803 | 36.0 | 7920 | 0.7253 | 0.7805 |
| 0.3045 | 37.0 | 8140 | 0.9133 | 0.6829 |
| 0.2202 | 38.0 | 8360 | 0.8750 | 0.7073 |
| 0.222 | 39.0 | 8580 | 0.5361 | 0.7805 |
| 0.1816 | 40.0 | 8800 | 1.1395 | 0.7317 |
| 0.2213 | 41.0 | 9020 | 0.9746 | 0.7561 |
| 0.16 | 42.0 | 9240 | 1.0585 | 0.7317 |
| 0.1084 | 43.0 | 9460 | 0.8736 | 0.7073 |
| 0.1243 | 44.0 | 9680 | 1.1686 | 0.8049 |
| 0.1009 | 45.0 | 9900 | 0.9862 | 0.7561 |
| 0.096 | 46.0 | 10120 | 1.4994 | 0.6829 |
| 0.089 | 47.0 | 10340 | 1.4415 | 0.7561 |
| 0.0473 | 48.0 | 10560 | 1.6197 | 0.7805 |
| 0.0266 | 49.0 | 10780 | 1.5765 | 0.7805 |
| 0.0418 | 50.0 | 11000 | 1.6080 | 0.7805 |
Framework versions
- Transformers 4.32.1
- Pytorch 2.1.0+cu121
- Datasets 2.12.0
- Tokenizers 0.13.2