vit-base-patch16-224-U8-40
This model is a fine-tuned version of google/vit-base-patch16-224 on the imagefolder dataset. It achieves the following results on the evaluation set:
- Loss: 0.5495
- Accuracy: 0.8667
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: 5.5e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.05
- num_epochs: 40
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 1.3457 | 1.0 | 20 | 1.3128 | 0.45 |
| 1.1498 | 2.0 | 40 | 1.1047 | 0.5667 |
| 0.8312 | 3.0 | 60 | 0.8231 | 0.65 |
| 0.5334 | 4.0 | 80 | 0.5719 | 0.8167 |
| 0.3582 | 5.0 | 100 | 0.5495 | 0.8667 |
| 0.2389 | 6.0 | 120 | 0.5801 | 0.8333 |
| 0.2055 | 7.0 | 140 | 0.6727 | 0.8167 |
| 0.1738 | 8.0 | 160 | 0.7238 | 0.8 |
| 0.1556 | 9.0 | 180 | 0.7665 | 0.75 |
| 0.1461 | 10.0 | 200 | 0.8229 | 0.7667 |
| 0.1401 | 11.0 | 220 | 0.8102 | 0.75 |
| 0.08 | 12.0 | 240 | 0.6609 | 0.8333 |
| 0.0989 | 13.0 | 260 | 0.6703 | 0.8333 |
| 0.0773 | 14.0 | 280 | 0.7303 | 0.8167 |
| 0.089 | 15.0 | 300 | 0.7757 | 0.7833 |
| 0.11 | 16.0 | 320 | 0.7279 | 0.8 |
| 0.086 | 17.0 | 340 | 0.8491 | 0.7833 |
| 0.0671 | 18.0 | 360 | 0.7950 | 0.8 |
| 0.0775 | 19.0 | 380 | 0.6753 | 0.85 |
| 0.0636 | 20.0 | 400 | 0.7881 | 0.8333 |
| 0.0737 | 21.0 | 420 | 0.7450 | 0.8333 |
| 0.0583 | 22.0 | 440 | 0.8295 | 0.8 |
| 0.0646 | 23.0 | 460 | 0.8227 | 0.8333 |
| 0.0637 | 24.0 | 480 | 0.9030 | 0.7833 |
| 0.0647 | 25.0 | 500 | 0.8656 | 0.8 |
| 0.0477 | 26.0 | 520 | 0.8362 | 0.8 |
| 0.0481 | 27.0 | 540 | 0.8389 | 0.8 |
| 0.0355 | 28.0 | 560 | 0.9424 | 0.8 |
| 0.0352 | 29.0 | 580 | 0.8963 | 0.8 |
| 0.0335 | 30.0 | 600 | 0.8560 | 0.8333 |
| 0.0372 | 31.0 | 620 | 0.7250 | 0.8333 |
| 0.0389 | 32.0 | 640 | 0.7846 | 0.8167 |
| 0.0425 | 33.0 | 660 | 0.8532 | 0.8333 |
| 0.0404 | 34.0 | 680 | 0.8169 | 0.8333 |
| 0.0359 | 35.0 | 700 | 0.8682 | 0.8167 |
| 0.0231 | 36.0 | 720 | 0.9362 | 0.8167 |
| 0.027 | 37.0 | 740 | 0.9139 | 0.8167 |
| 0.0214 | 38.0 | 760 | 0.8782 | 0.8167 |
| 0.0191 | 39.0 | 780 | 0.8794 | 0.8167 |
| 0.0293 | 40.0 | 800 | 0.8929 | 0.8167 |
Framework versions
- Transformers 4.36.2
- Pytorch 2.1.2+cu118
- Datasets 2.16.1
- Tokenizers 0.15.0