<!-- 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-tiny_rvl_cdip_100_examples_per_class_kd_MSE
This model is a fine-tuned version of WinKawaks/vit-tiny-patch16-224 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.7723
- Accuracy: 0.6025
- Brier Loss: 0.5295
- Nll: 3.6748
- F1 Micro: 0.6025
- F1 Macro: 0.6055
- Ece: 0.1688
- Aurc: 0.1708
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 | 4.7870 | 0.065 | 0.9655 | 17.0930 | 0.065 | 0.0550 | 0.1747 | 0.9357 |
No log | 2.0 | 50 | 3.9498 | 0.205 | 0.8858 | 9.5780 | 0.205 | 0.1863 | 0.1692 | 0.6618 |
No log | 3.0 | 75 | 3.3698 | 0.3675 | 0.7672 | 6.4908 | 0.3675 | 0.3392 | 0.1676 | 0.4195 |
No log | 4.0 | 100 | 2.9935 | 0.4075 | 0.6958 | 5.5595 | 0.4075 | 0.3820 | 0.1828 | 0.3327 |
No log | 5.0 | 125 | 2.8351 | 0.455 | 0.6591 | 4.8619 | 0.455 | 0.4351 | 0.1561 | 0.2833 |
No log | 6.0 | 150 | 2.8196 | 0.4725 | 0.6595 | 4.7785 | 0.4725 | 0.4367 | 0.1808 | 0.2790 |
No log | 7.0 | 175 | 2.6352 | 0.5075 | 0.6234 | 4.9881 | 0.5075 | 0.4886 | 0.1563 | 0.2493 |
No log | 8.0 | 200 | 2.5325 | 0.525 | 0.6162 | 4.3297 | 0.525 | 0.5026 | 0.1724 | 0.2365 |
No log | 9.0 | 225 | 2.5459 | 0.53 | 0.6099 | 5.1608 | 0.53 | 0.5148 | 0.1944 | 0.2350 |
No log | 10.0 | 250 | 2.5573 | 0.5325 | 0.6161 | 5.4495 | 0.5325 | 0.5212 | 0.2052 | 0.2397 |
No log | 11.0 | 275 | 2.3199 | 0.5675 | 0.5828 | 4.1247 | 0.5675 | 0.5626 | 0.1849 | 0.2071 |
No log | 12.0 | 300 | 2.2917 | 0.565 | 0.5758 | 4.1738 | 0.565 | 0.5694 | 0.1992 | 0.2023 |
No log | 13.0 | 325 | 2.2744 | 0.555 | 0.5974 | 4.2323 | 0.555 | 0.5544 | 0.1982 | 0.2203 |
No log | 14.0 | 350 | 2.1638 | 0.5625 | 0.5807 | 4.2049 | 0.5625 | 0.5629 | 0.1868 | 0.2049 |
No log | 15.0 | 375 | 2.1934 | 0.5575 | 0.5903 | 4.3813 | 0.5575 | 0.5614 | 0.1868 | 0.2022 |
No log | 16.0 | 400 | 2.1092 | 0.5625 | 0.5702 | 3.6094 | 0.5625 | 0.5700 | 0.1846 | 0.2011 |
No log | 17.0 | 425 | 2.0379 | 0.5875 | 0.5642 | 4.4351 | 0.5875 | 0.5822 | 0.2036 | 0.1959 |
No log | 18.0 | 450 | 2.0303 | 0.5825 | 0.5558 | 3.6847 | 0.5825 | 0.5820 | 0.1684 | 0.1881 |
No log | 19.0 | 475 | 2.0506 | 0.57 | 0.5749 | 4.0014 | 0.57 | 0.5708 | 0.1725 | 0.2027 |
1.5026 | 20.0 | 500 | 1.9932 | 0.5875 | 0.5524 | 3.8003 | 0.5875 | 0.5914 | 0.1843 | 0.1831 |
1.5026 | 21.0 | 525 | 2.0131 | 0.565 | 0.5643 | 4.0681 | 0.565 | 0.5635 | 0.1776 | 0.1957 |
1.5026 | 22.0 | 550 | 2.0162 | 0.5725 | 0.5712 | 3.7068 | 0.5725 | 0.5766 | 0.1934 | 0.1955 |
1.5026 | 23.0 | 575 | 1.9093 | 0.605 | 0.5381 | 3.7930 | 0.605 | 0.6032 | 0.1539 | 0.1749 |
1.5026 | 24.0 | 600 | 1.9607 | 0.575 | 0.5561 | 4.5740 | 0.575 | 0.5789 | 0.1782 | 0.1902 |
1.5026 | 25.0 | 625 | 1.8971 | 0.5825 | 0.5408 | 3.7290 | 0.5825 | 0.5754 | 0.1836 | 0.1751 |
1.5026 | 26.0 | 650 | 1.9217 | 0.5775 | 0.5537 | 3.8085 | 0.5775 | 0.5844 | 0.1725 | 0.1843 |
1.5026 | 27.0 | 675 | 1.9493 | 0.585 | 0.5606 | 3.6743 | 0.585 | 0.5953 | 0.1755 | 0.1882 |
1.5026 | 28.0 | 700 | 1.8884 | 0.585 | 0.5437 | 3.7865 | 0.585 | 0.5828 | 0.1801 | 0.1822 |
1.5026 | 29.0 | 725 | 1.9242 | 0.585 | 0.5479 | 3.9607 | 0.585 | 0.5856 | 0.1619 | 0.1817 |
1.5026 | 30.0 | 750 | 1.8767 | 0.5975 | 0.5470 | 3.7995 | 0.5975 | 0.5966 | 0.1599 | 0.1790 |
1.5026 | 31.0 | 775 | 1.8723 | 0.5925 | 0.5337 | 3.8962 | 0.5925 | 0.5972 | 0.1678 | 0.1729 |
1.5026 | 32.0 | 800 | 1.9093 | 0.585 | 0.5545 | 3.8776 | 0.585 | 0.5830 | 0.1902 | 0.1841 |
1.5026 | 33.0 | 825 | 1.8667 | 0.595 | 0.5363 | 3.8926 | 0.595 | 0.5917 | 0.1772 | 0.1745 |
1.5026 | 34.0 | 850 | 1.8403 | 0.59 | 0.5521 | 3.8560 | 0.59 | 0.5953 | 0.1711 | 0.1800 |
1.5026 | 35.0 | 875 | 1.8464 | 0.5925 | 0.5380 | 4.0376 | 0.5925 | 0.5970 | 0.1719 | 0.1756 |
1.5026 | 36.0 | 900 | 1.8441 | 0.5975 | 0.5411 | 3.7193 | 0.5975 | 0.6008 | 0.1569 | 0.1753 |
1.5026 | 37.0 | 925 | 1.8599 | 0.5875 | 0.5402 | 3.9139 | 0.5875 | 0.5908 | 0.1779 | 0.1789 |
1.5026 | 38.0 | 950 | 1.8559 | 0.6 | 0.5458 | 3.8970 | 0.6 | 0.5991 | 0.1583 | 0.1804 |
1.5026 | 39.0 | 975 | 1.8285 | 0.61 | 0.5370 | 3.6292 | 0.61 | 0.6155 | 0.1623 | 0.1722 |
0.0745 | 40.0 | 1000 | 1.8309 | 0.5975 | 0.5432 | 3.6865 | 0.5975 | 0.6017 | 0.1663 | 0.1821 |
0.0745 | 41.0 | 1025 | 1.8237 | 0.59 | 0.5348 | 3.6213 | 0.59 | 0.5921 | 0.1695 | 0.1738 |
0.0745 | 42.0 | 1050 | 1.8421 | 0.605 | 0.5360 | 3.8592 | 0.605 | 0.6048 | 0.1601 | 0.1743 |
0.0745 | 43.0 | 1075 | 1.8158 | 0.5975 | 0.5300 | 3.4537 | 0.5975 | 0.5953 | 0.1696 | 0.1707 |
0.0745 | 44.0 | 1100 | 1.8238 | 0.5875 | 0.5358 | 3.7706 | 0.5875 | 0.5923 | 0.1797 | 0.1754 |
0.0745 | 45.0 | 1125 | 1.8214 | 0.595 | 0.5463 | 3.4742 | 0.595 | 0.5981 | 0.1800 | 0.1770 |
0.0745 | 46.0 | 1150 | 1.8162 | 0.5925 | 0.5317 | 3.9260 | 0.5925 | 0.5950 | 0.1646 | 0.1733 |
0.0745 | 47.0 | 1175 | 1.8050 | 0.5975 | 0.5392 | 3.8322 | 0.5975 | 0.5979 | 0.1794 | 0.1763 |
0.0745 | 48.0 | 1200 | 1.8214 | 0.5975 | 0.5347 | 3.7965 | 0.5975 | 0.6009 | 0.1555 | 0.1746 |
0.0745 | 49.0 | 1225 | 1.7813 | 0.6 | 0.5294 | 3.8398 | 0.6 | 0.6005 | 0.1674 | 0.1688 |
0.0745 | 50.0 | 1250 | 1.8179 | 0.6075 | 0.5336 | 3.4690 | 0.6075 | 0.6112 | 0.1743 | 0.1748 |
0.0745 | 51.0 | 1275 | 1.7953 | 0.595 | 0.5380 | 3.7781 | 0.595 | 0.5990 | 0.1380 | 0.1727 |
0.0745 | 52.0 | 1300 | 1.7897 | 0.6 | 0.5323 | 3.7412 | 0.6 | 0.6013 | 0.1603 | 0.1707 |
0.0745 | 53.0 | 1325 | 1.8072 | 0.59 | 0.5428 | 3.5993 | 0.59 | 0.5947 | 0.1571 | 0.1773 |
0.0745 | 54.0 | 1350 | 1.7834 | 0.605 | 0.5219 | 3.7600 | 0.605 | 0.6049 | 0.1563 | 0.1671 |
0.0745 | 55.0 | 1375 | 1.7920 | 0.595 | 0.5361 | 3.5986 | 0.595 | 0.5978 | 0.1512 | 0.1717 |
0.0745 | 56.0 | 1400 | 1.8074 | 0.5925 | 0.5387 | 3.5383 | 0.5925 | 0.5962 | 0.1669 | 0.1741 |
0.0745 | 57.0 | 1425 | 1.7893 | 0.605 | 0.5346 | 3.6929 | 0.605 | 0.6039 | 0.1641 | 0.1681 |
0.0745 | 58.0 | 1450 | 1.7787 | 0.6 | 0.5317 | 3.7652 | 0.6 | 0.6004 | 0.1850 | 0.1726 |
0.0745 | 59.0 | 1475 | 1.7888 | 0.595 | 0.5323 | 3.4558 | 0.595 | 0.5975 | 0.1797 | 0.1732 |
0.0231 | 60.0 | 1500 | 1.8064 | 0.58 | 0.5332 | 3.7773 | 0.58 | 0.5839 | 0.1819 | 0.1762 |
0.0231 | 61.0 | 1525 | 1.7795 | 0.6075 | 0.5298 | 3.7998 | 0.6075 | 0.6086 | 0.1678 | 0.1704 |
0.0231 | 62.0 | 1550 | 1.7826 | 0.595 | 0.5318 | 3.6741 | 0.595 | 0.5916 | 0.1550 | 0.1715 |
0.0231 | 63.0 | 1575 | 1.7704 | 0.5925 | 0.5325 | 3.5942 | 0.5925 | 0.5941 | 0.1619 | 0.1712 |
0.0231 | 64.0 | 1600 | 1.7901 | 0.6025 | 0.5289 | 3.4459 | 0.6025 | 0.6054 | 0.2022 | 0.1712 |
0.0231 | 65.0 | 1625 | 1.7944 | 0.59 | 0.5381 | 3.7591 | 0.59 | 0.5910 | 0.1599 | 0.1756 |
0.0231 | 66.0 | 1650 | 1.7721 | 0.605 | 0.5256 | 3.5227 | 0.605 | 0.6045 | 0.1525 | 0.1677 |
0.0231 | 67.0 | 1675 | 1.7779 | 0.5975 | 0.5306 | 3.6792 | 0.5975 | 0.5994 | 0.1667 | 0.1714 |
0.0231 | 68.0 | 1700 | 1.7724 | 0.6 | 0.5250 | 3.7552 | 0.6 | 0.6022 | 0.1818 | 0.1683 |
0.0231 | 69.0 | 1725 | 1.7765 | 0.6025 | 0.5283 | 3.4264 | 0.6025 | 0.6019 | 0.1671 | 0.1700 |
0.0231 | 70.0 | 1750 | 1.7784 | 0.6 | 0.5276 | 3.6887 | 0.6 | 0.6053 | 0.1715 | 0.1703 |
0.0231 | 71.0 | 1775 | 1.7659 | 0.6 | 0.5282 | 3.6051 | 0.6 | 0.6006 | 0.1722 | 0.1691 |
0.0231 | 72.0 | 1800 | 1.7882 | 0.5975 | 0.5329 | 3.5950 | 0.5975 | 0.6016 | 0.1981 | 0.1716 |
0.0231 | 73.0 | 1825 | 1.7678 | 0.6 | 0.5287 | 3.6691 | 0.6 | 0.6032 | 0.1733 | 0.1696 |
0.0231 | 74.0 | 1850 | 1.7716 | 0.6 | 0.5286 | 3.7576 | 0.6 | 0.6013 | 0.1734 | 0.1692 |
0.0231 | 75.0 | 1875 | 1.7704 | 0.6 | 0.5299 | 3.5917 | 0.6 | 0.6016 | 0.1645 | 0.1709 |
0.0231 | 76.0 | 1900 | 1.7729 | 0.6 | 0.5298 | 3.6758 | 0.6 | 0.6024 | 0.1766 | 0.1710 |
0.0231 | 77.0 | 1925 | 1.7749 | 0.6 | 0.5308 | 3.6022 | 0.6 | 0.6030 | 0.1604 | 0.1717 |
0.0231 | 78.0 | 1950 | 1.7720 | 0.6 | 0.5294 | 3.6759 | 0.6 | 0.6017 | 0.1786 | 0.1708 |
0.0231 | 79.0 | 1975 | 1.7734 | 0.6025 | 0.5288 | 3.6765 | 0.6025 | 0.6048 | 0.1673 | 0.1698 |
0.0059 | 80.0 | 2000 | 1.7709 | 0.6 | 0.5286 | 3.6755 | 0.6 | 0.6020 | 0.1749 | 0.1704 |
0.0059 | 81.0 | 2025 | 1.7730 | 0.6 | 0.5295 | 3.6760 | 0.6 | 0.6020 | 0.1677 | 0.1708 |
0.0059 | 82.0 | 2050 | 1.7723 | 0.6025 | 0.5295 | 3.6756 | 0.6025 | 0.6055 | 0.1626 | 0.1708 |
0.0059 | 83.0 | 2075 | 1.7721 | 0.6025 | 0.5295 | 3.6741 | 0.6025 | 0.6055 | 0.1709 | 0.1708 |
0.0059 | 84.0 | 2100 | 1.7725 | 0.6025 | 0.5297 | 3.6747 | 0.6025 | 0.6048 | 0.1627 | 0.1709 |
0.0059 | 85.0 | 2125 | 1.7724 | 0.6025 | 0.5295 | 3.6751 | 0.6025 | 0.6055 | 0.1639 | 0.1707 |
0.0059 | 86.0 | 2150 | 1.7724 | 0.6025 | 0.5296 | 3.6751 | 0.6025 | 0.6055 | 0.1630 | 0.1708 |
0.0059 | 87.0 | 2175 | 1.7724 | 0.6025 | 0.5295 | 3.6749 | 0.6025 | 0.6055 | 0.1638 | 0.1707 |
0.0059 | 88.0 | 2200 | 1.7722 | 0.6025 | 0.5295 | 3.6752 | 0.6025 | 0.6055 | 0.1645 | 0.1708 |
0.0059 | 89.0 | 2225 | 1.7723 | 0.6025 | 0.5295 | 3.6747 | 0.6025 | 0.6055 | 0.1639 | 0.1708 |
0.0059 | 90.0 | 2250 | 1.7723 | 0.6025 | 0.5294 | 3.6750 | 0.6025 | 0.6055 | 0.1643 | 0.1708 |
0.0059 | 91.0 | 2275 | 1.7723 | 0.6025 | 0.5294 | 3.6750 | 0.6025 | 0.6055 | 0.1643 | 0.1708 |
0.0059 | 92.0 | 2300 | 1.7723 | 0.6025 | 0.5295 | 3.6747 | 0.6025 | 0.6055 | 0.1639 | 0.1708 |
0.0059 | 93.0 | 2325 | 1.7723 | 0.6025 | 0.5295 | 3.6749 | 0.6025 | 0.6055 | 0.1637 | 0.1707 |
0.0059 | 94.0 | 2350 | 1.7722 | 0.6025 | 0.5295 | 3.6749 | 0.6025 | 0.6055 | 0.1688 | 0.1708 |
0.0059 | 95.0 | 2375 | 1.7723 | 0.6025 | 0.5295 | 3.6748 | 0.6025 | 0.6055 | 0.1643 | 0.1708 |
0.0059 | 96.0 | 2400 | 1.7723 | 0.6025 | 0.5294 | 3.6748 | 0.6025 | 0.6055 | 0.1643 | 0.1707 |
0.0059 | 97.0 | 2425 | 1.7723 | 0.6025 | 0.5295 | 3.6748 | 0.6025 | 0.6055 | 0.1688 | 0.1708 |
0.0059 | 98.0 | 2450 | 1.7723 | 0.6025 | 0.5295 | 3.6749 | 0.6025 | 0.6055 | 0.1643 | 0.1708 |
0.0059 | 99.0 | 2475 | 1.7723 | 0.6025 | 0.5295 | 3.6749 | 0.6025 | 0.6055 | 0.1688 | 0.1708 |
0.0 | 100.0 | 2500 | 1.7723 | 0.6025 | 0.5295 | 3.6748 | 0.6025 | 0.6055 | 0.1688 | 0.1708 |
Framework versions
- Transformers 4.28.0.dev0
- Pytorch 1.12.1+cu113
- Datasets 2.12.0
- Tokenizers 0.12.1