hushem_40x_deit_tiny_adamax_0001_fold1
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: 1.3786
- Accuracy: 0.8444
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: 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 |
|---|---|---|---|---|
| 0.0587 | 1.0 | 215 | 0.8379 | 0.7778 |
| 0.0029 | 2.0 | 430 | 0.8134 | 0.8222 |
| 0.0061 | 3.0 | 645 | 0.6824 | 0.8667 |
| 0.0003 | 4.0 | 860 | 0.8964 | 0.8444 |
| 0.0004 | 5.0 | 1075 | 1.1389 | 0.8 |
| 0.0069 | 6.0 | 1290 | 0.8847 | 0.8222 |
| 0.0014 | 7.0 | 1505 | 0.9407 | 0.8444 |
| 0.0208 | 8.0 | 1720 | 1.2665 | 0.8 |
| 0.0 | 9.0 | 1935 | 0.7746 | 0.8222 |
| 0.0001 | 10.0 | 2150 | 0.9541 | 0.8222 |
| 0.0 | 11.0 | 2365 | 1.3297 | 0.7556 |
| 0.0 | 12.0 | 2580 | 1.2887 | 0.7778 |
| 0.0 | 13.0 | 2795 | 1.2405 | 0.7778 |
| 0.0 | 14.0 | 3010 | 1.2098 | 0.8 |
| 0.0 | 15.0 | 3225 | 1.1905 | 0.8 |
| 0.0 | 16.0 | 3440 | 1.1775 | 0.8 |
| 0.0 | 17.0 | 3655 | 1.1699 | 0.8 |
| 0.0 | 18.0 | 3870 | 1.1668 | 0.8 |
| 0.0 | 19.0 | 4085 | 1.1651 | 0.8 |
| 0.0 | 20.0 | 4300 | 1.1645 | 0.8 |
| 0.0 | 21.0 | 4515 | 1.1663 | 0.8 |
| 0.0 | 22.0 | 4730 | 1.1709 | 0.8 |
| 0.0 | 23.0 | 4945 | 1.1752 | 0.8 |
| 0.0 | 24.0 | 5160 | 1.1807 | 0.8 |
| 0.0 | 25.0 | 5375 | 1.1874 | 0.8222 |
| 0.0 | 26.0 | 5590 | 1.1925 | 0.8222 |
| 0.0 | 27.0 | 5805 | 1.1999 | 0.8222 |
| 0.0 | 28.0 | 6020 | 1.2057 | 0.8222 |
| 0.0 | 29.0 | 6235 | 1.2150 | 0.8222 |
| 0.0 | 30.0 | 6450 | 1.2228 | 0.8222 |
| 0.0 | 31.0 | 6665 | 1.2334 | 0.8222 |
| 0.0 | 32.0 | 6880 | 1.2399 | 0.8222 |
| 0.0 | 33.0 | 7095 | 1.2440 | 0.8222 |
| 0.0 | 34.0 | 7310 | 1.2539 | 0.8222 |
| 0.0 | 35.0 | 7525 | 1.2643 | 0.8222 |
| 0.0 | 36.0 | 7740 | 1.2752 | 0.8222 |
| 0.0 | 37.0 | 7955 | 1.2837 | 0.8222 |
| 0.0 | 38.0 | 8170 | 1.2941 | 0.8222 |
| 0.0 | 39.0 | 8385 | 1.3057 | 0.8444 |
| 0.0 | 40.0 | 8600 | 1.3171 | 0.8444 |
| 0.0 | 41.0 | 8815 | 1.3233 | 0.8444 |
| 0.0 | 42.0 | 9030 | 1.3334 | 0.8444 |
| 0.0 | 43.0 | 9245 | 1.3422 | 0.8444 |
| 0.0 | 44.0 | 9460 | 1.3487 | 0.8444 |
| 0.0 | 45.0 | 9675 | 1.3569 | 0.8444 |
| 0.0 | 46.0 | 9890 | 1.3629 | 0.8444 |
| 0.0 | 47.0 | 10105 | 1.3713 | 0.8444 |
| 0.0 | 48.0 | 10320 | 1.3761 | 0.8444 |
| 0.0 | 49.0 | 10535 | 1.3795 | 0.8444 |
| 0.0 | 50.0 | 10750 | 1.3786 | 0.8444 |
Framework versions
- Transformers 4.32.1
- Pytorch 2.1.1+cu121
- Datasets 2.12.0
- Tokenizers 0.13.2