vit-base_rvl-cdip-small_rvl_cdip-NK1000_og_simkd
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: 261.8253
- Accuracy: 0.845
- Brier Loss: 0.2896
- Nll: 1.8917
- F1 Micro: 0.845
- F1 Macro: 0.8458
- Ece: 0.1431
- Aurc: 0.0597
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 |
|---|---|---|---|---|---|---|---|---|---|---|
| 284.1528 | 1.0 | 1000 | 282.4832 | 0.5725 | 0.6206 | 2.4811 | 0.5725 | 0.5320 | 0.2175 | 0.1848 |
| 281.942 | 2.0 | 2000 | 280.8445 | 0.6943 | 0.4704 | 2.2781 | 0.6943 | 0.6823 | 0.1809 | 0.1131 |
| 281.4574 | 3.0 | 3000 | 280.4852 | 0.7185 | 0.4524 | 2.1204 | 0.7185 | 0.7250 | 0.1810 | 0.1008 |
| 279.8457 | 4.0 | 4000 | 278.7610 | 0.769 | 0.3964 | 2.0520 | 0.769 | 0.7685 | 0.1789 | 0.0775 |
| 279.1982 | 5.0 | 5000 | 278.0784 | 0.792 | 0.3522 | 1.9832 | 0.792 | 0.7915 | 0.1570 | 0.0670 |
| 278.1353 | 6.0 | 6000 | 277.1822 | 0.8135 | 0.3198 | 1.8943 | 0.8135 | 0.8160 | 0.1427 | 0.0547 |
| 277.4303 | 7.0 | 7000 | 275.9198 | 0.8193 | 0.3170 | 1.9321 | 0.8193 | 0.8203 | 0.1453 | 0.0589 |
| 276.2535 | 8.0 | 8000 | 274.8677 | 0.8273 | 0.3063 | 1.8543 | 0.8273 | 0.8266 | 0.1404 | 0.0538 |
| 275.1405 | 9.0 | 9000 | 273.8240 | 0.8345 | 0.2905 | 1.8312 | 0.8345 | 0.8362 | 0.1369 | 0.0525 |
| 274.3982 | 10.0 | 10000 | 273.2765 | 0.835 | 0.2892 | 1.8405 | 0.835 | 0.8362 | 0.1363 | 0.0512 |
| 272.9251 | 11.0 | 11000 | 272.4844 | 0.8455 | 0.2730 | 1.8874 | 0.8455 | 0.8468 | 0.1277 | 0.0478 |
| 272.1662 | 12.0 | 12000 | 271.4586 | 0.8373 | 0.2923 | 1.8514 | 0.8373 | 0.8374 | 0.1396 | 0.0508 |
| 272.1504 | 13.0 | 13000 | 271.0098 | 0.8452 | 0.2765 | 1.8428 | 0.8452 | 0.8454 | 0.1304 | 0.0505 |
| 271.0841 | 14.0 | 14000 | 270.4739 | 0.8405 | 0.2884 | 1.8279 | 0.8405 | 0.8421 | 0.1368 | 0.0522 |
| 270.5412 | 15.0 | 15000 | 269.5290 | 0.843 | 0.2861 | 1.8339 | 0.843 | 0.8434 | 0.1375 | 0.0524 |
| 269.4117 | 16.0 | 16000 | 269.1779 | 0.842 | 0.2874 | 1.8357 | 0.842 | 0.8422 | 0.1383 | 0.0520 |
| 269.1644 | 17.0 | 17000 | 268.5929 | 0.8465 | 0.2743 | 1.8563 | 0.8465 | 0.8470 | 0.1333 | 0.0491 |
| 268.7355 | 18.0 | 18000 | 268.2595 | 0.8475 | 0.2790 | 1.8540 | 0.8475 | 0.8479 | 0.1345 | 0.0505 |
| 268.3442 | 19.0 | 19000 | 267.7969 | 0.8508 | 0.2749 | 1.8406 | 0.8508 | 0.8509 | 0.1307 | 0.0505 |
| 267.4279 | 20.0 | 20000 | 267.2394 | 0.844 | 0.2811 | 1.8676 | 0.844 | 0.8448 | 0.1384 | 0.0509 |
| 267.468 | 21.0 | 21000 | 267.0267 | 0.8525 | 0.2694 | 1.8311 | 0.8525 | 0.8534 | 0.1293 | 0.0519 |
| 266.6685 | 22.0 | 22000 | 266.3500 | 0.8485 | 0.2772 | 1.8471 | 0.8485 | 0.8487 | 0.1368 | 0.0507 |
| 266.4612 | 23.0 | 23000 | 265.8022 | 0.8433 | 0.2863 | 1.8363 | 0.8433 | 0.8441 | 0.1399 | 0.0536 |
| 266.3148 | 24.0 | 24000 | 265.7575 | 0.8488 | 0.2783 | 1.8835 | 0.8488 | 0.8495 | 0.1366 | 0.0518 |
| 265.0058 | 25.0 | 25000 | 265.1237 | 0.8468 | 0.2841 | 1.8232 | 0.8468 | 0.8476 | 0.1370 | 0.0555 |
| 265.3975 | 26.0 | 26000 | 265.0540 | 0.8518 | 0.2757 | 1.8747 | 0.8518 | 0.8525 | 0.1324 | 0.0527 |
| 265.4347 | 27.0 | 27000 | 264.8875 | 0.8502 | 0.2755 | 1.8525 | 0.8502 | 0.8509 | 0.1339 | 0.0515 |
| 264.4956 | 28.0 | 28000 | 264.4421 | 0.8448 | 0.2864 | 1.8596 | 0.8448 | 0.8457 | 0.1402 | 0.0535 |
| 264.3941 | 29.0 | 29000 | 264.0486 | 0.8472 | 0.2815 | 1.8533 | 0.8472 | 0.8480 | 0.1379 | 0.0538 |
| 264.138 | 30.0 | 30000 | 264.2021 | 0.8495 | 0.2772 | 1.8547 | 0.8495 | 0.8500 | 0.1363 | 0.0531 |
| 263.8278 | 31.0 | 31000 | 263.6598 | 0.8472 | 0.2840 | 1.8715 | 0.8472 | 0.8472 | 0.1393 | 0.0549 |
| 263.4683 | 32.0 | 32000 | 263.4160 | 0.8465 | 0.2820 | 1.8844 | 0.8465 | 0.8471 | 0.1389 | 0.0558 |
| 263.5281 | 33.0 | 33000 | 263.2498 | 0.851 | 0.2788 | 1.8720 | 0.851 | 0.8520 | 0.1361 | 0.0554 |
| 263.3538 | 34.0 | 34000 | 262.9030 | 0.8472 | 0.2839 | 1.9007 | 0.8472 | 0.8482 | 0.1393 | 0.0562 |
| 262.673 | 35.0 | 35000 | 262.9031 | 0.8452 | 0.2859 | 1.8754 | 0.8452 | 0.8463 | 0.1406 | 0.0564 |
| 262.9104 | 36.0 | 36000 | 262.8404 | 0.8468 | 0.2867 | 1.8730 | 0.8468 | 0.8478 | 0.1398 | 0.0561 |
| 262.9824 | 37.0 | 37000 | 262.8044 | 0.849 | 0.2810 | 1.8759 | 0.849 | 0.8494 | 0.1372 | 0.0524 |
| 262.2614 | 38.0 | 38000 | 262.8396 | 0.8458 | 0.2861 | 1.8657 | 0.8458 | 0.8468 | 0.1410 | 0.0548 |
| 262.2726 | 39.0 | 39000 | 262.3623 | 0.846 | 0.2833 | 1.8772 | 0.8460 | 0.8465 | 0.1405 | 0.0565 |
| 262.3102 | 40.0 | 40000 | 262.4073 | 0.8465 | 0.2831 | 1.8798 | 0.8465 | 0.8475 | 0.1395 | 0.0553 |
| 262.2994 | 41.0 | 41000 | 262.2219 | 0.8472 | 0.2836 | 1.8810 | 0.8472 | 0.8475 | 0.1399 | 0.0579 |
| 262.222 | 42.0 | 42000 | 262.4181 | 0.8472 | 0.2775 | 1.8712 | 0.8472 | 0.8482 | 0.1389 | 0.0552 |
| 261.6536 | 43.0 | 43000 | 262.2162 | 0.8465 | 0.2844 | 1.8668 | 0.8465 | 0.8479 | 0.1401 | 0.0565 |
| 261.9964 | 44.0 | 44000 | 262.1039 | 0.8472 | 0.2848 | 1.8718 | 0.8472 | 0.8481 | 0.1403 | 0.0590 |
| 261.4522 | 45.0 | 45000 | 261.7883 | 0.846 | 0.2868 | 1.8589 | 0.8460 | 0.8459 | 0.1419 | 0.0556 |
| 261.6668 | 46.0 | 46000 | 262.0215 | 0.8492 | 0.2822 | 1.8682 | 0.8492 | 0.8494 | 0.1385 | 0.0542 |
| 261.8742 | 47.0 | 47000 | 261.9067 | 0.847 | 0.2846 | 1.8765 | 0.847 | 0.8476 | 0.1403 | 0.0599 |
| 261.5992 | 48.0 | 48000 | 261.7719 | 0.8475 | 0.2820 | 1.8854 | 0.8475 | 0.8485 | 0.1401 | 0.0583 |
| 261.6406 | 49.0 | 49000 | 261.5148 | 0.846 | 0.2873 | 1.8737 | 0.8460 | 0.8466 | 0.1427 | 0.0598 |
| 261.9611 | 50.0 | 50000 | 261.8253 | 0.845 | 0.2896 | 1.8917 | 0.845 | 0.8458 | 0.1431 | 0.0597 |
Framework versions
- Transformers 4.26.1
- Pytorch 1.13.1.post200
- Datasets 2.9.0
- Tokenizers 0.13.2