hushem_40x_deit_small_sgd_001_fold2
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: 0.9318
- Accuracy: 0.7333
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 |
|---|---|---|---|---|
| 1.1891 | 1.0 | 215 | 1.3300 | 0.3778 |
| 0.9647 | 2.0 | 430 | 1.2794 | 0.4444 |
| 0.8581 | 3.0 | 645 | 1.2244 | 0.5111 |
| 0.699 | 4.0 | 860 | 1.1784 | 0.5333 |
| 0.6158 | 5.0 | 1075 | 1.1498 | 0.5111 |
| 0.5391 | 6.0 | 1290 | 1.1059 | 0.5556 |
| 0.4953 | 7.0 | 1505 | 1.0650 | 0.5333 |
| 0.4016 | 8.0 | 1720 | 1.0249 | 0.5556 |
| 0.3397 | 9.0 | 1935 | 0.9796 | 0.6222 |
| 0.3003 | 10.0 | 2150 | 0.9463 | 0.7111 |
| 0.246 | 11.0 | 2365 | 0.9270 | 0.7111 |
| 0.1949 | 12.0 | 2580 | 0.9025 | 0.7111 |
| 0.1895 | 13.0 | 2795 | 0.8872 | 0.7111 |
| 0.1659 | 14.0 | 3010 | 0.8723 | 0.7111 |
| 0.1576 | 15.0 | 3225 | 0.8544 | 0.7111 |
| 0.1305 | 16.0 | 3440 | 0.8521 | 0.7111 |
| 0.1123 | 17.0 | 3655 | 0.8414 | 0.7111 |
| 0.1025 | 18.0 | 3870 | 0.8453 | 0.7111 |
| 0.0749 | 19.0 | 4085 | 0.8597 | 0.7111 |
| 0.0854 | 20.0 | 4300 | 0.8467 | 0.7111 |
| 0.0788 | 21.0 | 4515 | 0.8314 | 0.7111 |
| 0.0675 | 22.0 | 4730 | 0.8392 | 0.7111 |
| 0.0523 | 23.0 | 4945 | 0.8293 | 0.7111 |
| 0.0556 | 24.0 | 5160 | 0.8555 | 0.7111 |
| 0.0483 | 25.0 | 5375 | 0.8566 | 0.7111 |
| 0.0417 | 26.0 | 5590 | 0.8533 | 0.7111 |
| 0.0397 | 27.0 | 5805 | 0.8560 | 0.7333 |
| 0.0302 | 28.0 | 6020 | 0.8587 | 0.7333 |
| 0.0286 | 29.0 | 6235 | 0.8633 | 0.7333 |
| 0.0386 | 30.0 | 6450 | 0.8691 | 0.7333 |
| 0.0212 | 31.0 | 6665 | 0.8693 | 0.7333 |
| 0.0221 | 32.0 | 6880 | 0.8714 | 0.7333 |
| 0.0198 | 33.0 | 7095 | 0.8818 | 0.7333 |
| 0.0189 | 34.0 | 7310 | 0.8880 | 0.7333 |
| 0.0167 | 35.0 | 7525 | 0.8939 | 0.7333 |
| 0.0198 | 36.0 | 7740 | 0.9010 | 0.7333 |
| 0.0157 | 37.0 | 7955 | 0.8988 | 0.7333 |
| 0.0177 | 38.0 | 8170 | 0.9154 | 0.7333 |
| 0.0136 | 39.0 | 8385 | 0.9094 | 0.7333 |
| 0.0108 | 40.0 | 8600 | 0.9213 | 0.7333 |
| 0.0119 | 41.0 | 8815 | 0.9173 | 0.7333 |
| 0.0127 | 42.0 | 9030 | 0.9219 | 0.7333 |
| 0.0095 | 43.0 | 9245 | 0.9256 | 0.7333 |
| 0.0124 | 44.0 | 9460 | 0.9223 | 0.7333 |
| 0.0112 | 45.0 | 9675 | 0.9246 | 0.7333 |
| 0.0112 | 46.0 | 9890 | 0.9266 | 0.7333 |
| 0.0102 | 47.0 | 10105 | 0.9301 | 0.7333 |
| 0.0105 | 48.0 | 10320 | 0.9338 | 0.7333 |
| 0.0119 | 49.0 | 10535 | 0.9314 | 0.7333 |
| 0.0144 | 50.0 | 10750 | 0.9318 | 0.7333 |
Framework versions
- Transformers 4.32.1
- Pytorch 2.1.0+cu121
- Datasets 2.12.0
- Tokenizers 0.13.2