hushem_40x_deit_tiny_rms_001_fold4
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: 3.6233
- Accuracy: 0.6905
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 |
|---|---|---|---|---|
| 1.1473 | 1.0 | 219 | 0.8756 | 0.6190 |
| 0.9719 | 2.0 | 438 | 0.9893 | 0.5714 |
| 0.7611 | 3.0 | 657 | 0.7217 | 0.7619 |
| 0.6995 | 4.0 | 876 | 0.7516 | 0.6429 |
| 0.6928 | 5.0 | 1095 | 1.0447 | 0.5952 |
| 0.6114 | 6.0 | 1314 | 0.9410 | 0.6667 |
| 0.4906 | 7.0 | 1533 | 1.4457 | 0.5238 |
| 0.4956 | 8.0 | 1752 | 1.1229 | 0.6429 |
| 0.3708 | 9.0 | 1971 | 0.5610 | 0.7381 |
| 0.3213 | 10.0 | 2190 | 1.1632 | 0.6667 |
| 0.279 | 11.0 | 2409 | 0.8853 | 0.7381 |
| 0.26 | 12.0 | 2628 | 1.0316 | 0.6905 |
| 0.2004 | 13.0 | 2847 | 0.8001 | 0.7619 |
| 0.2396 | 14.0 | 3066 | 1.0495 | 0.7381 |
| 0.1937 | 15.0 | 3285 | 1.2736 | 0.7381 |
| 0.1386 | 16.0 | 3504 | 0.9949 | 0.7381 |
| 0.1459 | 17.0 | 3723 | 1.2302 | 0.6905 |
| 0.0754 | 18.0 | 3942 | 1.9238 | 0.6667 |
| 0.0996 | 19.0 | 4161 | 1.4396 | 0.6905 |
| 0.0438 | 20.0 | 4380 | 1.1891 | 0.7143 |
| 0.1349 | 21.0 | 4599 | 1.4228 | 0.7381 |
| 0.0058 | 22.0 | 4818 | 1.2340 | 0.7619 |
| 0.0345 | 23.0 | 5037 | 1.1630 | 0.6667 |
| 0.0461 | 24.0 | 5256 | 2.1318 | 0.6429 |
| 0.0595 | 25.0 | 5475 | 1.7499 | 0.6905 |
| 0.004 | 26.0 | 5694 | 1.6488 | 0.6905 |
| 0.0014 | 27.0 | 5913 | 1.8134 | 0.6905 |
| 0.0335 | 28.0 | 6132 | 2.3351 | 0.6905 |
| 0.0071 | 29.0 | 6351 | 2.4170 | 0.5714 |
| 0.0006 | 30.0 | 6570 | 1.5965 | 0.7381 |
| 0.0014 | 31.0 | 6789 | 2.0937 | 0.7381 |
| 0.0419 | 32.0 | 7008 | 1.8845 | 0.6905 |
| 0.042 | 33.0 | 7227 | 3.6234 | 0.6190 |
| 0.0018 | 34.0 | 7446 | 1.5177 | 0.6905 |
| 0.027 | 35.0 | 7665 | 1.3824 | 0.7857 |
| 0.0005 | 36.0 | 7884 | 2.3915 | 0.7619 |
| 0.0226 | 37.0 | 8103 | 1.6001 | 0.7381 |
| 0.001 | 38.0 | 8322 | 2.3141 | 0.6905 |
| 0.0 | 39.0 | 8541 | 2.5460 | 0.7143 |
| 0.0 | 40.0 | 8760 | 2.6724 | 0.6905 |
| 0.0 | 41.0 | 8979 | 2.7005 | 0.6905 |
| 0.0 | 42.0 | 9198 | 2.8171 | 0.7143 |
| 0.0 | 43.0 | 9417 | 2.9876 | 0.7143 |
| 0.0 | 44.0 | 9636 | 3.1125 | 0.7143 |
| 0.0 | 45.0 | 9855 | 3.2479 | 0.7143 |
| 0.0 | 46.0 | 10074 | 3.4344 | 0.7143 |
| 0.0 | 47.0 | 10293 | 3.4573 | 0.7143 |
| 0.0 | 48.0 | 10512 | 3.5752 | 0.6905 |
| 0.0 | 49.0 | 10731 | 3.5910 | 0.6905 |
| 0.0 | 50.0 | 10950 | 3.6233 | 0.6905 |
Framework versions
- Transformers 4.32.1
- Pytorch 2.1.1+cu121
- Datasets 2.12.0
- Tokenizers 0.13.2