<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
vit-base-patch16-224-in21k-small_rvl_cdip-NK1000_hint
This model is a fine-tuned version of WinKawaks/vit-small-patch16-224 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 75.2362
- Accuracy: 0.829
- Brier Loss: 0.3192
- Nll: 1.9987
- F1 Micro: 0.8290
- F1 Macro: 0.8288
- Ece: 0.1536
- Aurc: 0.0562
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: 16
- eval_batch_size: 16
- 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 | Brier Loss | Nll | F1 Micro | F1 Macro | Ece | Aurc |
|---|---|---|---|---|---|---|---|---|---|---|
| 78.2197 | 1.0 | 1000 | 77.9227 | 0.6322 | 0.4884 | 2.2807 | 0.6322 | 0.6226 | 0.0652 | 0.1538 |
| 77.1947 | 2.0 | 2000 | 77.1839 | 0.727 | 0.3752 | 2.1097 | 0.7270 | 0.7290 | 0.0480 | 0.0917 |
| 76.83 | 3.0 | 3000 | 76.9771 | 0.7385 | 0.3662 | 2.0823 | 0.7385 | 0.7421 | 0.0612 | 0.0827 |
| 76.6666 | 4.0 | 4000 | 76.7764 | 0.7725 | 0.3383 | 1.9757 | 0.7725 | 0.7763 | 0.0808 | 0.0697 |
| 76.2784 | 5.0 | 5000 | 76.6028 | 0.773 | 0.3271 | 2.0074 | 0.7730 | 0.7737 | 0.0708 | 0.0668 |
| 75.9985 | 6.0 | 6000 | 76.4455 | 0.7875 | 0.3138 | 1.9622 | 0.7875 | 0.7874 | 0.0868 | 0.0593 |
| 75.7339 | 7.0 | 7000 | 76.2669 | 0.7973 | 0.3127 | 1.9661 | 0.7973 | 0.7978 | 0.1123 | 0.0546 |
| 75.504 | 8.0 | 8000 | 76.3972 | 0.7927 | 0.3373 | 2.0115 | 0.7927 | 0.7954 | 0.1318 | 0.0594 |
| 75.3558 | 9.0 | 9000 | 76.4614 | 0.7785 | 0.3628 | 2.1225 | 0.7785 | 0.7793 | 0.1501 | 0.0648 |
| 75.1485 | 10.0 | 10000 | 76.2131 | 0.795 | 0.3351 | 2.0123 | 0.795 | 0.7964 | 0.1467 | 0.0565 |
| 75.1098 | 11.0 | 11000 | 76.3161 | 0.798 | 0.3452 | 2.0430 | 0.798 | 0.8002 | 0.1569 | 0.0587 |
| 74.822 | 12.0 | 12000 | 76.2536 | 0.788 | 0.3637 | 2.0509 | 0.788 | 0.7877 | 0.1659 | 0.0607 |
| 74.8787 | 13.0 | 13000 | 76.2025 | 0.7965 | 0.3493 | 2.0401 | 0.7965 | 0.7981 | 0.1538 | 0.0582 |
| 74.7046 | 14.0 | 14000 | 76.1598 | 0.8075 | 0.3335 | 2.0388 | 0.8075 | 0.8058 | 0.1551 | 0.0525 |
| 74.6157 | 15.0 | 15000 | 76.0894 | 0.8003 | 0.3496 | 1.9931 | 0.8003 | 0.8006 | 0.1615 | 0.0566 |
| 74.6451 | 16.0 | 16000 | 76.0082 | 0.8065 | 0.3380 | 2.0005 | 0.8065 | 0.8060 | 0.1593 | 0.0530 |
| 74.3042 | 17.0 | 17000 | 76.0281 | 0.8075 | 0.3398 | 2.0028 | 0.8075 | 0.8097 | 0.1592 | 0.0544 |
| 74.3261 | 18.0 | 18000 | 75.9836 | 0.8063 | 0.3414 | 2.0447 | 0.8062 | 0.8066 | 0.1614 | 0.0566 |
| 74.1196 | 19.0 | 19000 | 75.8935 | 0.8103 | 0.3347 | 2.0211 | 0.8103 | 0.8121 | 0.1592 | 0.0595 |
| 74.2291 | 20.0 | 20000 | 75.9679 | 0.815 | 0.3329 | 2.0335 | 0.815 | 0.8141 | 0.1586 | 0.0572 |
| 74.268 | 21.0 | 21000 | 76.0052 | 0.8073 | 0.3442 | 2.0847 | 0.8073 | 0.8067 | 0.1633 | 0.0589 |
| 74.0436 | 22.0 | 22000 | 75.9529 | 0.8093 | 0.3454 | 2.1010 | 0.8093 | 0.8081 | 0.1625 | 0.0547 |
| 73.9289 | 23.0 | 23000 | 75.8841 | 0.8103 | 0.3420 | 2.0569 | 0.8103 | 0.8104 | 0.1625 | 0.0580 |
| 73.9519 | 24.0 | 24000 | 75.7295 | 0.8167 | 0.3320 | 2.0459 | 0.8167 | 0.8152 | 0.1575 | 0.0533 |
| 73.9333 | 25.0 | 25000 | 75.6503 | 0.8165 | 0.3296 | 1.9681 | 0.8165 | 0.8174 | 0.1586 | 0.0523 |
| 73.8239 | 26.0 | 26000 | 75.6156 | 0.8203 | 0.3245 | 2.0540 | 0.8203 | 0.8192 | 0.1546 | 0.0506 |
| 73.7011 | 27.0 | 27000 | 75.7075 | 0.8183 | 0.3312 | 2.0996 | 0.8183 | 0.8193 | 0.1594 | 0.0562 |
| 73.4822 | 28.0 | 28000 | 75.5065 | 0.8247 | 0.3184 | 2.0404 | 0.8247 | 0.8254 | 0.1535 | 0.0548 |
| 73.5787 | 29.0 | 29000 | 75.6063 | 0.8193 | 0.3295 | 2.0527 | 0.8193 | 0.8192 | 0.1591 | 0.0560 |
| 73.519 | 30.0 | 30000 | 75.5828 | 0.8163 | 0.3351 | 2.0151 | 0.8163 | 0.8173 | 0.1621 | 0.0582 |
| 73.6516 | 31.0 | 31000 | 75.4986 | 0.827 | 0.3147 | 2.0640 | 0.827 | 0.8272 | 0.1513 | 0.0539 |
| 73.5156 | 32.0 | 32000 | 75.5884 | 0.8147 | 0.3355 | 2.0634 | 0.8148 | 0.8137 | 0.1631 | 0.0556 |
| 73.4564 | 33.0 | 33000 | 75.3992 | 0.8233 | 0.3219 | 2.0498 | 0.8233 | 0.8227 | 0.1536 | 0.0526 |
| 73.3286 | 34.0 | 34000 | 75.4277 | 0.8197 | 0.3256 | 2.0222 | 0.8197 | 0.8213 | 0.1594 | 0.0540 |
| 73.3056 | 35.0 | 35000 | 75.3989 | 0.8285 | 0.3136 | 1.9681 | 0.8285 | 0.8303 | 0.1510 | 0.0566 |
| 73.3272 | 36.0 | 36000 | 75.4398 | 0.8233 | 0.3247 | 2.0504 | 0.8233 | 0.8247 | 0.1583 | 0.0553 |
| 73.2738 | 37.0 | 37000 | 75.3631 | 0.8207 | 0.3242 | 1.9921 | 0.8207 | 0.8211 | 0.1595 | 0.0546 |
| 73.2657 | 38.0 | 38000 | 75.3613 | 0.8245 | 0.3231 | 2.0715 | 0.8245 | 0.8232 | 0.1569 | 0.0548 |
| 73.2045 | 39.0 | 39000 | 75.3697 | 0.8223 | 0.3253 | 2.0207 | 0.8223 | 0.8213 | 0.1571 | 0.0557 |
| 73.1701 | 40.0 | 40000 | 75.3138 | 0.8277 | 0.3174 | 2.0071 | 0.8277 | 0.8282 | 0.1525 | 0.0557 |
| 73.1491 | 41.0 | 41000 | 75.3160 | 0.827 | 0.3183 | 2.0131 | 0.827 | 0.8271 | 0.1549 | 0.0573 |
| 73.1466 | 42.0 | 42000 | 75.3052 | 0.8297 | 0.3166 | 1.9978 | 0.8297 | 0.8296 | 0.1535 | 0.0558 |
| 73.0658 | 43.0 | 43000 | 75.3064 | 0.8293 | 0.3166 | 2.0293 | 0.8293 | 0.8292 | 0.1548 | 0.0570 |
| 73.1394 | 44.0 | 44000 | 75.2527 | 0.8285 | 0.3179 | 2.0172 | 0.8285 | 0.8284 | 0.1540 | 0.0554 |
| 73.2385 | 45.0 | 45000 | 75.2782 | 0.828 | 0.3181 | 2.0026 | 0.828 | 0.8280 | 0.1556 | 0.0570 |
| 73.2207 | 46.0 | 46000 | 75.2624 | 0.827 | 0.3194 | 1.9884 | 0.827 | 0.8268 | 0.1552 | 0.0559 |
| 73.2837 | 47.0 | 47000 | 75.2604 | 0.8285 | 0.3195 | 1.9982 | 0.8285 | 0.8283 | 0.1542 | 0.0566 |
| 73.0848 | 48.0 | 48000 | 75.2454 | 0.829 | 0.3188 | 1.9958 | 0.8290 | 0.8288 | 0.1535 | 0.0568 |
| 73.111 | 49.0 | 49000 | 75.2438 | 0.8283 | 0.3196 | 2.0019 | 0.8283 | 0.8282 | 0.1542 | 0.0567 |
| 73.0278 | 50.0 | 50000 | 75.2362 | 0.829 | 0.3192 | 1.9987 | 0.8290 | 0.8288 | 0.1536 | 0.0562 |
Framework versions
- Transformers 4.26.1
- Pytorch 1.13.1.post200
- Datasets 2.9.0
- Tokenizers 0.13.2
