<!-- 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_rvl-cdip-tiny_rvl_cdip-NK1000_kd
This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.8222
- Accuracy: 0.5543
- Brier Loss: 0.6966
- Nll: 3.2790
- F1 Micro: 0.5543
- F1 Macro: 0.5553
- Ece: 0.2764
- Aurc: 0.2323
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: 128
- eval_batch_size: 128
- 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 |
|---|---|---|---|---|---|---|---|---|---|---|
| No log | 1.0 | 125 | 2.4938 | 0.1348 | 0.9105 | 6.1860 | 0.1348 | 0.0673 | 0.0701 | 0.7303 |
| No log | 2.0 | 250 | 2.3746 | 0.1895 | 0.8922 | 5.4000 | 0.1895 | 0.1384 | 0.0653 | 0.6997 |
| No log | 3.0 | 375 | 2.2780 | 0.198 | 0.8746 | 4.3716 | 0.198 | 0.1340 | 0.0676 | 0.6597 |
| 2.396 | 4.0 | 500 | 2.0622 | 0.302 | 0.8212 | 4.2504 | 0.302 | 0.2268 | 0.0655 | 0.5280 |
| 2.396 | 5.0 | 625 | 1.8529 | 0.3703 | 0.7693 | 3.3328 | 0.3703 | 0.3184 | 0.0727 | 0.4470 |
| 2.396 | 6.0 | 750 | 1.6847 | 0.423 | 0.7103 | 3.0730 | 0.4230 | 0.3879 | 0.0698 | 0.3567 |
| 2.396 | 7.0 | 875 | 1.5975 | 0.45 | 0.6817 | 3.0713 | 0.45 | 0.4139 | 0.0632 | 0.3257 |
| 1.7095 | 8.0 | 1000 | 1.5156 | 0.4768 | 0.6588 | 2.8252 | 0.4768 | 0.4517 | 0.0635 | 0.3029 |
| 1.7095 | 9.0 | 1125 | 1.4425 | 0.5018 | 0.6308 | 2.7656 | 0.5018 | 0.4812 | 0.0650 | 0.2728 |
| 1.7095 | 10.0 | 1250 | 1.4089 | 0.5092 | 0.6218 | 2.6715 | 0.5092 | 0.4894 | 0.0527 | 0.2642 |
| 1.7095 | 11.0 | 1375 | 1.3930 | 0.523 | 0.6150 | 2.6821 | 0.523 | 0.5261 | 0.0635 | 0.2584 |
| 1.3064 | 12.0 | 1500 | 1.4166 | 0.5205 | 0.6262 | 2.7691 | 0.5205 | 0.4991 | 0.0813 | 0.2639 |
| 1.3064 | 13.0 | 1625 | 1.3343 | 0.5312 | 0.5961 | 2.6475 | 0.5312 | 0.5194 | 0.0586 | 0.2383 |
| 1.3064 | 14.0 | 1750 | 1.3277 | 0.5417 | 0.5917 | 2.6528 | 0.5417 | 0.5361 | 0.0669 | 0.2327 |
| 1.3064 | 15.0 | 1875 | 1.3407 | 0.5312 | 0.5958 | 2.6880 | 0.5312 | 0.5356 | 0.0637 | 0.2378 |
| 1.0419 | 16.0 | 2000 | 1.2873 | 0.5545 | 0.5801 | 2.6042 | 0.5545 | 0.5509 | 0.0870 | 0.2193 |
| 1.0419 | 17.0 | 2125 | 1.3539 | 0.5375 | 0.6022 | 2.6706 | 0.5375 | 0.5329 | 0.0970 | 0.2376 |
| 1.0419 | 18.0 | 2250 | 1.3073 | 0.5543 | 0.5857 | 2.6217 | 0.5543 | 0.5502 | 0.1006 | 0.2200 |
| 1.0419 | 19.0 | 2375 | 1.3225 | 0.558 | 0.5886 | 2.6258 | 0.558 | 0.5530 | 0.1047 | 0.2206 |
| 0.8001 | 20.0 | 2500 | 1.3573 | 0.554 | 0.5955 | 2.7139 | 0.554 | 0.5489 | 0.1221 | 0.2200 |
| 0.8001 | 21.0 | 2625 | 1.4029 | 0.546 | 0.6150 | 2.7649 | 0.546 | 0.5456 | 0.1547 | 0.2274 |
| 0.8001 | 22.0 | 2750 | 1.4006 | 0.5525 | 0.6092 | 2.8131 | 0.5525 | 0.5504 | 0.1474 | 0.2246 |
| 0.8001 | 23.0 | 2875 | 1.4523 | 0.5513 | 0.6223 | 2.8803 | 0.5513 | 0.5448 | 0.1818 | 0.2269 |
| 0.5716 | 24.0 | 3000 | 1.4744 | 0.5495 | 0.6261 | 2.9958 | 0.5495 | 0.5525 | 0.1799 | 0.2253 |
| 0.5716 | 25.0 | 3125 | 1.5278 | 0.5445 | 0.6418 | 3.0853 | 0.5445 | 0.5485 | 0.1915 | 0.2321 |
| 0.5716 | 26.0 | 3250 | 1.5782 | 0.5433 | 0.6566 | 3.0618 | 0.5433 | 0.5448 | 0.2171 | 0.2333 |
| 0.5716 | 27.0 | 3375 | 1.6368 | 0.5375 | 0.6704 | 3.2249 | 0.5375 | 0.5389 | 0.2277 | 0.2401 |
| 0.3744 | 28.0 | 3500 | 1.6339 | 0.5445 | 0.6694 | 3.1689 | 0.5445 | 0.5447 | 0.2376 | 0.2338 |
| 0.3744 | 29.0 | 3625 | 1.6589 | 0.548 | 0.6714 | 3.1654 | 0.548 | 0.5469 | 0.2376 | 0.2319 |
| 0.3744 | 30.0 | 3750 | 1.7679 | 0.5353 | 0.6989 | 3.3537 | 0.5353 | 0.5387 | 0.2524 | 0.2558 |
| 0.3744 | 31.0 | 3875 | 1.7441 | 0.5475 | 0.6846 | 3.3716 | 0.5475 | 0.5501 | 0.2455 | 0.2395 |
| 0.2439 | 32.0 | 4000 | 1.7856 | 0.5365 | 0.6977 | 3.4176 | 0.5365 | 0.5443 | 0.2510 | 0.2462 |
| 0.2439 | 33.0 | 4125 | 1.7886 | 0.545 | 0.6997 | 3.3804 | 0.545 | 0.5454 | 0.2646 | 0.2379 |
| 0.2439 | 34.0 | 4250 | 1.8658 | 0.5275 | 0.7187 | 3.6006 | 0.5275 | 0.5300 | 0.2840 | 0.2482 |
| 0.2439 | 35.0 | 4375 | 1.8668 | 0.5387 | 0.7145 | 3.3922 | 0.5387 | 0.5391 | 0.2797 | 0.2453 |
| 0.1695 | 36.0 | 4500 | 1.8920 | 0.5288 | 0.7263 | 3.4756 | 0.5288 | 0.5320 | 0.2878 | 0.2507 |
| 0.1695 | 37.0 | 4625 | 1.8767 | 0.542 | 0.7146 | 3.5924 | 0.542 | 0.5357 | 0.2792 | 0.2469 |
| 0.1695 | 38.0 | 4750 | 1.8617 | 0.5435 | 0.7094 | 3.5434 | 0.5435 | 0.5467 | 0.2729 | 0.2440 |
| 0.1695 | 39.0 | 4875 | 1.8746 | 0.5525 | 0.7073 | 3.4325 | 0.5525 | 0.5514 | 0.2789 | 0.2434 |
| 0.1278 | 40.0 | 5000 | 1.8877 | 0.5435 | 0.7171 | 3.4872 | 0.5435 | 0.5438 | 0.2852 | 0.2393 |
| 0.1278 | 41.0 | 5125 | 1.8919 | 0.54 | 0.7219 | 3.4577 | 0.54 | 0.5456 | 0.2869 | 0.2487 |
| 0.1278 | 42.0 | 5250 | 1.8631 | 0.548 | 0.7089 | 3.4287 | 0.548 | 0.5502 | 0.2758 | 0.2390 |
| 0.1278 | 43.0 | 5375 | 1.8433 | 0.5475 | 0.7058 | 3.2993 | 0.5475 | 0.5468 | 0.2863 | 0.2335 |
| 0.0993 | 44.0 | 5500 | 1.8458 | 0.5505 | 0.7048 | 3.3852 | 0.5505 | 0.5528 | 0.2776 | 0.2378 |
| 0.0993 | 45.0 | 5625 | 1.8408 | 0.5443 | 0.7100 | 3.3510 | 0.5443 | 0.5490 | 0.2769 | 0.2392 |
| 0.0993 | 46.0 | 5750 | 1.8492 | 0.5477 | 0.7064 | 3.2989 | 0.5477 | 0.5496 | 0.2807 | 0.2363 |
| 0.0993 | 47.0 | 5875 | 1.8100 | 0.5497 | 0.6969 | 3.2853 | 0.5497 | 0.5534 | 0.2761 | 0.2341 |
| 0.0803 | 48.0 | 6000 | 1.8260 | 0.5523 | 0.6984 | 3.2543 | 0.5523 | 0.5532 | 0.2783 | 0.2326 |
| 0.0803 | 49.0 | 6125 | 1.8225 | 0.5563 | 0.6970 | 3.3070 | 0.5563 | 0.5573 | 0.2739 | 0.2327 |
| 0.0803 | 50.0 | 6250 | 1.8222 | 0.5543 | 0.6966 | 3.2790 | 0.5543 | 0.5553 | 0.2764 | 0.2323 |
Framework versions
- Transformers 4.26.1
- Pytorch 1.13.1.post200
- Datasets 2.9.0
- Tokenizers 0.13.2
