hushem_40x_deit_small_sgd_0001_fold1
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: 1.3807
- Accuracy: 0.3556
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 |
|---|---|---|---|---|
| 1.7636 | 1.0 | 215 | 1.5271 | 0.2222 |
| 1.5603 | 2.0 | 430 | 1.4695 | 0.2222 |
| 1.4189 | 3.0 | 645 | 1.4485 | 0.2444 |
| 1.4216 | 4.0 | 860 | 1.4404 | 0.3333 |
| 1.3718 | 5.0 | 1075 | 1.4361 | 0.3333 |
| 1.3271 | 6.0 | 1290 | 1.4331 | 0.3556 |
| 1.3291 | 7.0 | 1505 | 1.4309 | 0.3556 |
| 1.2611 | 8.0 | 1720 | 1.4294 | 0.3333 |
| 1.2392 | 9.0 | 1935 | 1.4281 | 0.3333 |
| 1.2352 | 10.0 | 2150 | 1.4273 | 0.3778 |
| 1.2132 | 11.0 | 2365 | 1.4268 | 0.3778 |
| 1.17 | 12.0 | 2580 | 1.4262 | 0.3778 |
| 1.1599 | 13.0 | 2795 | 1.4258 | 0.3778 |
| 1.1465 | 14.0 | 3010 | 1.4259 | 0.3778 |
| 1.1384 | 15.0 | 3225 | 1.4258 | 0.3556 |
| 1.1196 | 16.0 | 3440 | 1.4258 | 0.3333 |
| 1.1235 | 17.0 | 3655 | 1.4254 | 0.3333 |
| 1.092 | 18.0 | 3870 | 1.4252 | 0.3333 |
| 1.0493 | 19.0 | 4085 | 1.4248 | 0.3333 |
| 1.0602 | 20.0 | 4300 | 1.4241 | 0.2889 |
| 1.0537 | 21.0 | 4515 | 1.4232 | 0.2889 |
| 1.0424 | 22.0 | 4730 | 1.4223 | 0.2889 |
| 1.0373 | 23.0 | 4945 | 1.4208 | 0.2889 |
| 1.0255 | 24.0 | 5160 | 1.4191 | 0.3111 |
| 0.9946 | 25.0 | 5375 | 1.4173 | 0.3111 |
| 0.9526 | 26.0 | 5590 | 1.4155 | 0.3111 |
| 0.961 | 27.0 | 5805 | 1.4133 | 0.3111 |
| 0.9603 | 28.0 | 6020 | 1.4115 | 0.3111 |
| 0.9689 | 29.0 | 6235 | 1.4091 | 0.3111 |
| 0.9155 | 30.0 | 6450 | 1.4068 | 0.3111 |
| 0.9244 | 31.0 | 6665 | 1.4046 | 0.3111 |
| 0.9454 | 32.0 | 6880 | 1.4024 | 0.3111 |
| 0.9669 | 33.0 | 7095 | 1.4003 | 0.3111 |
| 0.935 | 34.0 | 7310 | 1.3982 | 0.3333 |
| 0.887 | 35.0 | 7525 | 1.3962 | 0.3333 |
| 0.9142 | 36.0 | 7740 | 1.3943 | 0.3333 |
| 0.9282 | 37.0 | 7955 | 1.3924 | 0.3333 |
| 0.8935 | 38.0 | 8170 | 1.3908 | 0.3333 |
| 0.9345 | 39.0 | 8385 | 1.3890 | 0.3333 |
| 0.8406 | 40.0 | 8600 | 1.3876 | 0.3333 |
| 0.8885 | 41.0 | 8815 | 1.3862 | 0.3333 |
| 0.9974 | 42.0 | 9030 | 1.3851 | 0.3333 |
| 0.9464 | 43.0 | 9245 | 1.3840 | 0.3333 |
| 0.9071 | 44.0 | 9460 | 1.3830 | 0.3333 |
| 0.9277 | 45.0 | 9675 | 1.3823 | 0.3333 |
| 0.8844 | 46.0 | 9890 | 1.3817 | 0.3333 |
| 0.8843 | 47.0 | 10105 | 1.3812 | 0.3333 |
| 0.9119 | 48.0 | 10320 | 1.3809 | 0.3556 |
| 0.9448 | 49.0 | 10535 | 1.3808 | 0.3556 |
| 0.8919 | 50.0 | 10750 | 1.3807 | 0.3556 |
Framework versions
- Transformers 4.32.1
- Pytorch 2.1.0+cu121
- Datasets 2.12.0
- Tokenizers 0.13.2