hushem_5x_deit_tiny_adamax_00001_fold5
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: 1.0196
- Accuracy: 0.8293
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 |
|---|---|---|---|---|
| 1.3464 | 1.0 | 28 | 1.2206 | 0.4390 |
| 1.038 | 2.0 | 56 | 1.1163 | 0.4390 |
| 0.9036 | 3.0 | 84 | 1.0326 | 0.5122 |
| 0.7256 | 4.0 | 112 | 0.9850 | 0.4634 |
| 0.6091 | 5.0 | 140 | 0.9299 | 0.5366 |
| 0.5118 | 6.0 | 168 | 0.8096 | 0.6098 |
| 0.3976 | 7.0 | 196 | 0.8337 | 0.6341 |
| 0.2983 | 8.0 | 224 | 0.8361 | 0.6829 |
| 0.2464 | 9.0 | 252 | 0.7489 | 0.6829 |
| 0.1797 | 10.0 | 280 | 0.7126 | 0.7317 |
| 0.155 | 11.0 | 308 | 0.7190 | 0.7317 |
| 0.1087 | 12.0 | 336 | 0.7349 | 0.7561 |
| 0.0817 | 13.0 | 364 | 0.6756 | 0.7805 |
| 0.0808 | 14.0 | 392 | 0.7587 | 0.7561 |
| 0.0526 | 15.0 | 420 | 0.6534 | 0.7805 |
| 0.0415 | 16.0 | 448 | 0.7396 | 0.7805 |
| 0.0249 | 17.0 | 476 | 0.7772 | 0.8049 |
| 0.0224 | 18.0 | 504 | 0.7783 | 0.8049 |
| 0.016 | 19.0 | 532 | 0.8153 | 0.7805 |
| 0.0121 | 20.0 | 560 | 0.8052 | 0.8049 |
| 0.0082 | 21.0 | 588 | 0.8047 | 0.8049 |
| 0.0059 | 22.0 | 616 | 0.8544 | 0.8049 |
| 0.0042 | 23.0 | 644 | 0.9271 | 0.7805 |
| 0.0032 | 24.0 | 672 | 0.8999 | 0.8049 |
| 0.0029 | 25.0 | 700 | 0.9068 | 0.8293 |
| 0.0025 | 26.0 | 728 | 0.9094 | 0.8293 |
| 0.0022 | 27.0 | 756 | 0.9291 | 0.8293 |
| 0.0019 | 28.0 | 784 | 0.9347 | 0.8293 |
| 0.0016 | 29.0 | 812 | 0.9448 | 0.8293 |
| 0.0016 | 30.0 | 840 | 0.9586 | 0.8293 |
| 0.0015 | 31.0 | 868 | 0.9704 | 0.8293 |
| 0.0013 | 32.0 | 896 | 0.9735 | 0.8293 |
| 0.0013 | 33.0 | 924 | 0.9776 | 0.8293 |
| 0.0012 | 34.0 | 952 | 0.9829 | 0.8293 |
| 0.0011 | 35.0 | 980 | 0.9923 | 0.8293 |
| 0.0011 | 36.0 | 1008 | 0.9922 | 0.8293 |
| 0.001 | 37.0 | 1036 | 0.9983 | 0.8293 |
| 0.001 | 38.0 | 1064 | 1.0035 | 0.8293 |
| 0.0009 | 39.0 | 1092 | 0.9985 | 0.8293 |
| 0.0009 | 40.0 | 1120 | 1.0064 | 0.8293 |
| 0.0009 | 41.0 | 1148 | 1.0089 | 0.8293 |
| 0.0008 | 42.0 | 1176 | 1.0130 | 0.8293 |
| 0.0008 | 43.0 | 1204 | 1.0152 | 0.8293 |
| 0.0009 | 44.0 | 1232 | 1.0185 | 0.8293 |
| 0.0009 | 45.0 | 1260 | 1.0165 | 0.8293 |
| 0.0009 | 46.0 | 1288 | 1.0180 | 0.8293 |
| 0.0008 | 47.0 | 1316 | 1.0182 | 0.8293 |
| 0.0008 | 48.0 | 1344 | 1.0196 | 0.8293 |
| 0.0008 | 49.0 | 1372 | 1.0196 | 0.8293 |
| 0.0008 | 50.0 | 1400 | 1.0196 | 0.8293 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu118
- Datasets 2.15.0
- Tokenizers 0.15.0