<!-- 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-small_rvl_cdip_100_examples_per_class_kd_CEKD_t1.5_a0.9
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: 1.2366
- Accuracy: 0.63
- Brier Loss: 0.5035
- Nll: 2.8588
- F1 Micro: 0.63
- F1 Macro: 0.6311
- Ece: 0.1649
- Aurc: 0.1472
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: 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: 100
Training results
Training Loss | Epoch | Step | Validation Loss | Accuracy | Brier Loss | Nll | F1 Micro | F1 Macro | Ece | Aurc |
---|---|---|---|---|---|---|---|---|---|---|
No log | 1.0 | 25 | 2.8887 | 0.1225 | 0.9306 | 15.9457 | 0.1225 | 0.1226 | 0.1434 | 0.8620 |
No log | 2.0 | 50 | 2.2120 | 0.3775 | 0.7577 | 9.7500 | 0.3775 | 0.3483 | 0.1992 | 0.3776 |
No log | 3.0 | 75 | 1.7681 | 0.495 | 0.6387 | 5.6935 | 0.495 | 0.4838 | 0.1885 | 0.2491 |
No log | 4.0 | 100 | 1.6420 | 0.5225 | 0.6038 | 5.2427 | 0.5225 | 0.5242 | 0.1757 | 0.2301 |
No log | 5.0 | 125 | 1.5877 | 0.545 | 0.5986 | 4.6187 | 0.545 | 0.5282 | 0.1808 | 0.2248 |
No log | 6.0 | 150 | 1.6460 | 0.5125 | 0.6162 | 3.9942 | 0.5125 | 0.5060 | 0.1962 | 0.2295 |
No log | 7.0 | 175 | 1.8436 | 0.5125 | 0.6538 | 4.1740 | 0.5125 | 0.4932 | 0.2299 | 0.2451 |
No log | 8.0 | 200 | 1.8205 | 0.545 | 0.6453 | 5.0752 | 0.545 | 0.5234 | 0.2057 | 0.2432 |
No log | 9.0 | 225 | 1.7399 | 0.55 | 0.6260 | 4.5896 | 0.55 | 0.5460 | 0.2057 | 0.2258 |
No log | 10.0 | 250 | 1.8559 | 0.55 | 0.6521 | 5.0532 | 0.55 | 0.5368 | 0.2209 | 0.2560 |
No log | 11.0 | 275 | 1.8636 | 0.5625 | 0.6488 | 4.6642 | 0.5625 | 0.5544 | 0.2335 | 0.2187 |
No log | 12.0 | 300 | 1.7461 | 0.55 | 0.6356 | 4.1298 | 0.55 | 0.5638 | 0.2047 | 0.2313 |
No log | 13.0 | 325 | 1.7468 | 0.5625 | 0.6281 | 4.5451 | 0.5625 | 0.5570 | 0.2224 | 0.2214 |
No log | 14.0 | 350 | 1.9616 | 0.545 | 0.6884 | 3.7999 | 0.545 | 0.5484 | 0.2691 | 0.2624 |
No log | 15.0 | 375 | 2.0977 | 0.5175 | 0.7138 | 4.3792 | 0.5175 | 0.5055 | 0.2658 | 0.2917 |
No log | 16.0 | 400 | 2.0238 | 0.5275 | 0.6896 | 4.5299 | 0.5275 | 0.5177 | 0.2664 | 0.2603 |
No log | 17.0 | 425 | 1.8687 | 0.535 | 0.6534 | 3.7356 | 0.535 | 0.5388 | 0.2490 | 0.2448 |
No log | 18.0 | 450 | 1.8210 | 0.5575 | 0.6492 | 4.3823 | 0.5575 | 0.5537 | 0.2533 | 0.2268 |
No log | 19.0 | 475 | 1.7610 | 0.555 | 0.6325 | 3.9697 | 0.555 | 0.5503 | 0.2292 | 0.2161 |
0.5398 | 20.0 | 500 | 1.7125 | 0.5825 | 0.6125 | 3.4176 | 0.5825 | 0.5731 | 0.2140 | 0.1859 |
0.5398 | 21.0 | 525 | 1.6296 | 0.5775 | 0.6163 | 3.6014 | 0.5775 | 0.5871 | 0.2236 | 0.2051 |
0.5398 | 22.0 | 550 | 1.5965 | 0.57 | 0.5908 | 3.7668 | 0.57 | 0.5712 | 0.2058 | 0.1883 |
0.5398 | 23.0 | 575 | 1.4828 | 0.5875 | 0.5646 | 3.7028 | 0.5875 | 0.5854 | 0.1944 | 0.1714 |
0.5398 | 24.0 | 600 | 1.3983 | 0.6075 | 0.5481 | 3.3608 | 0.6075 | 0.6107 | 0.1966 | 0.1628 |
0.5398 | 25.0 | 625 | 1.5241 | 0.5925 | 0.5866 | 3.3669 | 0.5925 | 0.6019 | 0.2069 | 0.1886 |
0.5398 | 26.0 | 650 | 1.5540 | 0.58 | 0.5780 | 3.5184 | 0.58 | 0.5710 | 0.2131 | 0.1857 |
0.5398 | 27.0 | 675 | 1.4653 | 0.6 | 0.5768 | 2.9877 | 0.6 | 0.6043 | 0.2166 | 0.1781 |
0.5398 | 28.0 | 700 | 1.4883 | 0.5925 | 0.5646 | 3.7789 | 0.5925 | 0.5910 | 0.2096 | 0.1746 |
0.5398 | 29.0 | 725 | 1.5738 | 0.59 | 0.5914 | 4.0558 | 0.59 | 0.5879 | 0.2150 | 0.1957 |
0.5398 | 30.0 | 750 | 1.4017 | 0.6025 | 0.5583 | 3.4791 | 0.6025 | 0.6023 | 0.2150 | 0.1752 |
0.5398 | 31.0 | 775 | 1.3500 | 0.61 | 0.5365 | 3.2560 | 0.61 | 0.6157 | 0.1988 | 0.1579 |
0.5398 | 32.0 | 800 | 1.2977 | 0.6375 | 0.5140 | 3.0503 | 0.6375 | 0.6395 | 0.1847 | 0.1534 |
0.5398 | 33.0 | 825 | 1.3471 | 0.6175 | 0.5406 | 3.1888 | 0.6175 | 0.6104 | 0.2077 | 0.1689 |
0.5398 | 34.0 | 850 | 1.2992 | 0.615 | 0.5219 | 2.8944 | 0.615 | 0.6191 | 0.1826 | 0.1574 |
0.5398 | 35.0 | 875 | 1.2733 | 0.6225 | 0.5124 | 2.9352 | 0.6225 | 0.6238 | 0.1588 | 0.1505 |
0.5398 | 36.0 | 900 | 1.2821 | 0.6175 | 0.5231 | 3.0142 | 0.6175 | 0.6169 | 0.1672 | 0.1553 |
0.5398 | 37.0 | 925 | 1.2819 | 0.61 | 0.5200 | 2.6874 | 0.61 | 0.6116 | 0.1847 | 0.1540 |
0.5398 | 38.0 | 950 | 1.2664 | 0.615 | 0.5145 | 2.9287 | 0.615 | 0.6159 | 0.1961 | 0.1528 |
0.5398 | 39.0 | 975 | 1.2584 | 0.6225 | 0.5134 | 3.0058 | 0.6225 | 0.6230 | 0.1747 | 0.1508 |
0.0507 | 40.0 | 1000 | 1.2562 | 0.615 | 0.5114 | 2.9269 | 0.615 | 0.6169 | 0.1815 | 0.1504 |
0.0507 | 41.0 | 1025 | 1.2525 | 0.6225 | 0.5101 | 2.9199 | 0.6225 | 0.6239 | 0.1770 | 0.1496 |
0.0507 | 42.0 | 1050 | 1.2573 | 0.62 | 0.5133 | 2.9195 | 0.62 | 0.6221 | 0.1824 | 0.1511 |
0.0507 | 43.0 | 1075 | 1.2536 | 0.6125 | 0.5131 | 2.9026 | 0.6125 | 0.6121 | 0.1820 | 0.1511 |
0.0507 | 44.0 | 1100 | 1.2543 | 0.6225 | 0.5109 | 3.0693 | 0.6225 | 0.6235 | 0.1647 | 0.1500 |
0.0507 | 45.0 | 1125 | 1.2526 | 0.6125 | 0.5117 | 2.9018 | 0.6125 | 0.6141 | 0.1788 | 0.1500 |
0.0507 | 46.0 | 1150 | 1.2432 | 0.615 | 0.5068 | 2.9042 | 0.615 | 0.6167 | 0.1762 | 0.1484 |
0.0507 | 47.0 | 1175 | 1.2485 | 0.6275 | 0.5098 | 2.8927 | 0.6275 | 0.6251 | 0.1590 | 0.1496 |
0.0507 | 48.0 | 1200 | 1.2576 | 0.6125 | 0.5140 | 2.8956 | 0.6125 | 0.6137 | 0.1824 | 0.1524 |
0.0507 | 49.0 | 1225 | 1.2468 | 0.62 | 0.5094 | 2.8918 | 0.62 | 0.6204 | 0.1832 | 0.1496 |
0.0507 | 50.0 | 1250 | 1.2479 | 0.6175 | 0.5102 | 2.8921 | 0.6175 | 0.6178 | 0.1706 | 0.1491 |
0.0507 | 51.0 | 1275 | 1.2393 | 0.6225 | 0.5057 | 2.8813 | 0.6225 | 0.6229 | 0.1784 | 0.1486 |
0.0507 | 52.0 | 1300 | 1.2463 | 0.6175 | 0.5085 | 2.8959 | 0.6175 | 0.6184 | 0.1669 | 0.1495 |
0.0507 | 53.0 | 1325 | 1.2391 | 0.62 | 0.5061 | 2.8828 | 0.62 | 0.6215 | 0.1803 | 0.1471 |
0.0507 | 54.0 | 1350 | 1.2538 | 0.6175 | 0.5121 | 2.8795 | 0.6175 | 0.6167 | 0.1680 | 0.1512 |
0.0507 | 55.0 | 1375 | 1.2407 | 0.625 | 0.5064 | 2.8830 | 0.625 | 0.6259 | 0.1842 | 0.1482 |
0.0507 | 56.0 | 1400 | 1.2488 | 0.62 | 0.5099 | 2.8769 | 0.62 | 0.6198 | 0.1568 | 0.1499 |
0.0507 | 57.0 | 1425 | 1.2402 | 0.625 | 0.5052 | 2.8778 | 0.625 | 0.6260 | 0.1616 | 0.1481 |
0.0507 | 58.0 | 1450 | 1.2457 | 0.625 | 0.5077 | 2.8786 | 0.625 | 0.6260 | 0.1759 | 0.1474 |
0.0507 | 59.0 | 1475 | 1.2430 | 0.6275 | 0.5073 | 2.8744 | 0.6275 | 0.6266 | 0.1652 | 0.1486 |
0.0319 | 60.0 | 1500 | 1.2399 | 0.625 | 0.5056 | 2.8767 | 0.625 | 0.6256 | 0.1701 | 0.1474 |
0.0319 | 61.0 | 1525 | 1.2460 | 0.63 | 0.5087 | 2.8758 | 0.63 | 0.6329 | 0.1865 | 0.1491 |
0.0319 | 62.0 | 1550 | 1.2410 | 0.6225 | 0.5058 | 2.8719 | 0.6225 | 0.6229 | 0.1752 | 0.1477 |
0.0319 | 63.0 | 1575 | 1.2418 | 0.63 | 0.5060 | 2.8746 | 0.63 | 0.6319 | 0.1692 | 0.1484 |
0.0319 | 64.0 | 1600 | 1.2424 | 0.6275 | 0.5069 | 2.8672 | 0.6275 | 0.6279 | 0.1903 | 0.1475 |
0.0319 | 65.0 | 1625 | 1.2413 | 0.63 | 0.5061 | 2.8747 | 0.63 | 0.6304 | 0.1737 | 0.1471 |
0.0319 | 66.0 | 1650 | 1.2385 | 0.6325 | 0.5039 | 2.8726 | 0.6325 | 0.6358 | 0.1792 | 0.1473 |
0.0319 | 67.0 | 1675 | 1.2368 | 0.625 | 0.5047 | 2.8661 | 0.625 | 0.6261 | 0.1843 | 0.1467 |
0.0319 | 68.0 | 1700 | 1.2370 | 0.6275 | 0.5039 | 2.8691 | 0.6275 | 0.6294 | 0.1724 | 0.1471 |
0.0319 | 69.0 | 1725 | 1.2382 | 0.63 | 0.5050 | 2.8659 | 0.63 | 0.6317 | 0.1698 | 0.1472 |
0.0319 | 70.0 | 1750 | 1.2396 | 0.6275 | 0.5051 | 2.8670 | 0.6275 | 0.6290 | 0.1790 | 0.1474 |
0.0319 | 71.0 | 1775 | 1.2378 | 0.625 | 0.5045 | 2.8637 | 0.625 | 0.6268 | 0.1742 | 0.1476 |
0.0319 | 72.0 | 1800 | 1.2360 | 0.625 | 0.5037 | 2.8669 | 0.625 | 0.6269 | 0.1778 | 0.1468 |
0.0319 | 73.0 | 1825 | 1.2390 | 0.63 | 0.5049 | 2.8638 | 0.63 | 0.6310 | 0.1711 | 0.1474 |
0.0319 | 74.0 | 1850 | 1.2372 | 0.625 | 0.5045 | 2.8640 | 0.625 | 0.6269 | 0.1817 | 0.1475 |
0.0319 | 75.0 | 1875 | 1.2375 | 0.63 | 0.5044 | 2.8640 | 0.63 | 0.6313 | 0.1703 | 0.1472 |
0.0319 | 76.0 | 1900 | 1.2372 | 0.6275 | 0.5041 | 2.8621 | 0.6275 | 0.6290 | 0.1794 | 0.1473 |
0.0319 | 77.0 | 1925 | 1.2374 | 0.63 | 0.5041 | 2.8629 | 0.63 | 0.6313 | 0.1722 | 0.1472 |
0.0319 | 78.0 | 1950 | 1.2367 | 0.6275 | 0.5039 | 2.8620 | 0.6275 | 0.6294 | 0.1704 | 0.1474 |
0.0319 | 79.0 | 1975 | 1.2371 | 0.6275 | 0.5039 | 2.8619 | 0.6275 | 0.6294 | 0.1639 | 0.1474 |
0.0314 | 80.0 | 2000 | 1.2372 | 0.63 | 0.5041 | 2.8612 | 0.63 | 0.6310 | 0.1750 | 0.1474 |
0.0314 | 81.0 | 2025 | 1.2368 | 0.63 | 0.5038 | 2.8613 | 0.63 | 0.6309 | 0.1648 | 0.1473 |
0.0314 | 82.0 | 2050 | 1.2370 | 0.63 | 0.5038 | 2.8607 | 0.63 | 0.6305 | 0.1782 | 0.1473 |
0.0314 | 83.0 | 2075 | 1.2368 | 0.63 | 0.5038 | 2.8609 | 0.63 | 0.6307 | 0.1686 | 0.1472 |
0.0314 | 84.0 | 2100 | 1.2368 | 0.63 | 0.5037 | 2.8603 | 0.63 | 0.6305 | 0.1667 | 0.1472 |
0.0314 | 85.0 | 2125 | 1.2366 | 0.63 | 0.5036 | 2.8601 | 0.63 | 0.6309 | 0.1686 | 0.1473 |
0.0314 | 86.0 | 2150 | 1.2367 | 0.6325 | 0.5037 | 2.8600 | 0.6325 | 0.6335 | 0.1751 | 0.1471 |
0.0314 | 87.0 | 2175 | 1.2369 | 0.63 | 0.5037 | 2.8598 | 0.63 | 0.6307 | 0.1730 | 0.1473 |
0.0314 | 88.0 | 2200 | 1.2367 | 0.63 | 0.5036 | 2.8595 | 0.63 | 0.6307 | 0.1657 | 0.1472 |
0.0314 | 89.0 | 2225 | 1.2366 | 0.63 | 0.5036 | 2.8597 | 0.63 | 0.6307 | 0.1680 | 0.1472 |
0.0314 | 90.0 | 2250 | 1.2366 | 0.63 | 0.5036 | 2.8594 | 0.63 | 0.6307 | 0.1580 | 0.1472 |
0.0314 | 91.0 | 2275 | 1.2366 | 0.63 | 0.5035 | 2.8593 | 0.63 | 0.6307 | 0.1677 | 0.1472 |
0.0314 | 92.0 | 2300 | 1.2367 | 0.63 | 0.5035 | 2.8593 | 0.63 | 0.6307 | 0.1616 | 0.1472 |
0.0314 | 93.0 | 2325 | 1.2366 | 0.63 | 0.5035 | 2.8590 | 0.63 | 0.6307 | 0.1625 | 0.1472 |
0.0314 | 94.0 | 2350 | 1.2366 | 0.6325 | 0.5035 | 2.8590 | 0.6325 | 0.6333 | 0.1586 | 0.1470 |
0.0314 | 95.0 | 2375 | 1.2366 | 0.63 | 0.5035 | 2.8591 | 0.63 | 0.6307 | 0.1580 | 0.1472 |
0.0314 | 96.0 | 2400 | 1.2366 | 0.63 | 0.5035 | 2.8589 | 0.63 | 0.6307 | 0.1695 | 0.1471 |
0.0314 | 97.0 | 2425 | 1.2366 | 0.63 | 0.5035 | 2.8589 | 0.63 | 0.6311 | 0.1648 | 0.1472 |
0.0314 | 98.0 | 2450 | 1.2366 | 0.63 | 0.5035 | 2.8588 | 0.63 | 0.6311 | 0.1695 | 0.1471 |
0.0314 | 99.0 | 2475 | 1.2366 | 0.6325 | 0.5035 | 2.8589 | 0.6325 | 0.6337 | 0.1724 | 0.1470 |
0.0312 | 100.0 | 2500 | 1.2366 | 0.63 | 0.5035 | 2.8588 | 0.63 | 0.6311 | 0.1649 | 0.1472 |
Framework versions
- Transformers 4.28.0.dev0
- Pytorch 1.12.1+cu113
- Datasets 2.12.0
- Tokenizers 0.12.1