Edit model card

dit-base_tobacco-tiny_tobacco3482_hint

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: 2.1196
  • Accuracy: 0.805
  • Brier Loss: 0.3299
  • Nll: 1.3687
  • F1 Micro: 0.805
  • F1 Macro: 0.7917
  • Ece: 0.1606
  • Aurc: 0.0741

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: 100

Training results

Training Loss Epoch Step Validation Loss Accuracy Brier Loss Nll F1 Micro F1 Macro Ece Aurc
No log 1.0 7 3.7183 0.225 0.9191 7.9210 0.225 0.1037 0.3197 0.7539
No log 2.0 14 3.2602 0.365 0.7808 4.4817 0.3650 0.2558 0.2908 0.5116
No log 3.0 21 2.8475 0.53 0.6305 2.7026 0.53 0.4263 0.2496 0.2812
No log 4.0 28 2.6118 0.575 0.5382 1.8007 0.575 0.5478 0.2435 0.2022
No log 5.0 35 2.4289 0.66 0.4404 1.6436 0.66 0.6317 0.2124 0.1236
No log 6.0 42 2.4007 0.69 0.4123 1.9255 0.69 0.6896 0.2038 0.1195
No log 7.0 49 2.3239 0.745 0.3775 1.9428 0.745 0.7365 0.1838 0.0957
No log 8.0 56 2.3109 0.735 0.3660 1.6472 0.735 0.7370 0.1729 0.0842
No log 9.0 63 2.3263 0.74 0.3867 1.5935 0.74 0.7422 0.1803 0.0944
No log 10.0 70 2.3878 0.735 0.4037 1.6604 0.735 0.7039 0.2057 0.0909
No log 11.0 77 2.5009 0.715 0.4220 1.8526 0.715 0.6918 0.2184 0.1017
No log 12.0 84 2.5428 0.72 0.4169 1.8799 0.72 0.6894 0.2000 0.0979
No log 13.0 91 2.5197 0.765 0.3884 1.8802 0.765 0.7268 0.1895 0.0933
No log 14.0 98 2.4928 0.725 0.4147 1.6096 0.7250 0.6923 0.2078 0.0940
No log 15.0 105 2.4172 0.765 0.3748 1.6754 0.765 0.7317 0.1874 0.0824
No log 16.0 112 2.4116 0.76 0.3873 1.4155 0.76 0.7280 0.1972 0.0890
No log 17.0 119 2.9196 0.73 0.4500 1.8493 0.7300 0.7220 0.2290 0.1657
No log 18.0 126 2.5815 0.765 0.4175 2.0326 0.765 0.7467 0.2125 0.0990
No log 19.0 133 2.7076 0.735 0.4475 1.8526 0.735 0.7152 0.2062 0.1049
No log 20.0 140 2.6951 0.71 0.4709 2.0521 0.7100 0.7258 0.2380 0.1188
No log 21.0 147 2.4037 0.765 0.4013 1.8740 0.765 0.7691 0.2009 0.0949
No log 22.0 154 2.6585 0.73 0.4303 2.0299 0.7300 0.7170 0.2110 0.1004
No log 23.0 161 2.4320 0.75 0.3950 1.8720 0.75 0.7340 0.2004 0.0895
No log 24.0 168 2.4891 0.74 0.4199 1.6458 0.74 0.7458 0.2227 0.1128
No log 25.0 175 2.6550 0.705 0.4833 1.7755 0.705 0.7042 0.2478 0.1145
No log 26.0 182 2.3191 0.765 0.3965 1.6941 0.765 0.7415 0.1954 0.0731
No log 27.0 189 2.3000 0.785 0.3763 1.3143 0.785 0.7416 0.2026 0.0712
No log 28.0 196 2.2047 0.78 0.3409 1.4818 0.78 0.7694 0.1759 0.0688
No log 29.0 203 2.3587 0.77 0.3781 1.6779 0.7700 0.7571 0.1937 0.0766
No log 30.0 210 2.5027 0.75 0.4400 1.8454 0.75 0.7338 0.2232 0.0817
No log 31.0 217 2.4092 0.77 0.3899 1.5010 0.7700 0.7498 0.1987 0.0710
No log 32.0 224 2.7655 0.74 0.4520 2.0720 0.74 0.7177 0.2266 0.0855
No log 33.0 231 2.3814 0.76 0.3979 1.4053 0.76 0.7352 0.1982 0.0754
No log 34.0 238 2.3946 0.775 0.3790 1.6969 0.775 0.7387 0.1908 0.0876
No log 35.0 245 2.5158 0.775 0.4064 1.4329 0.775 0.7428 0.2068 0.0929
No log 36.0 252 2.4920 0.75 0.4281 1.5724 0.75 0.7470 0.2161 0.1028
No log 37.0 259 2.4541 0.765 0.3842 1.5272 0.765 0.7163 0.1918 0.0853
No log 38.0 266 2.3785 0.82 0.3319 1.6360 0.82 0.7944 0.1732 0.0811
No log 39.0 273 2.3721 0.77 0.3831 1.2724 0.7700 0.7649 0.1880 0.0818
No log 40.0 280 2.5684 0.795 0.3494 1.7291 0.795 0.7724 0.1771 0.1126
No log 41.0 287 2.4835 0.78 0.3799 1.7700 0.78 0.7594 0.2024 0.0821
No log 42.0 294 2.4690 0.795 0.3678 1.6379 0.795 0.7804 0.1836 0.0840
No log 43.0 301 2.3069 0.77 0.3809 1.4970 0.7700 0.7621 0.1807 0.0784
No log 44.0 308 2.4424 0.795 0.3558 1.7162 0.795 0.7857 0.1879 0.0667
No log 45.0 315 2.0018 0.86 0.2387 1.4924 0.8600 0.8558 0.1180 0.0526
No log 46.0 322 2.4174 0.785 0.3593 1.7688 0.785 0.7662 0.1859 0.0773
No log 47.0 329 2.1816 0.84 0.2862 1.5093 0.8400 0.8230 0.1459 0.0679
No log 48.0 336 2.3006 0.78 0.3601 1.7675 0.78 0.7496 0.1847 0.0752
No log 49.0 343 2.2952 0.81 0.3231 1.4851 0.81 0.7865 0.1691 0.0656
No log 50.0 350 2.1346 0.8 0.3132 1.4625 0.8000 0.7937 0.1660 0.0789
No log 51.0 357 2.2935 0.81 0.3383 1.5764 0.81 0.7914 0.1770 0.0717
No log 52.0 364 2.1792 0.825 0.3116 1.5652 0.825 0.8163 0.1454 0.0654
No log 53.0 371 2.1231 0.81 0.3066 1.3012 0.81 0.8062 0.1552 0.0604
No log 54.0 378 1.9712 0.825 0.2854 1.2891 0.825 0.8137 0.1430 0.0521
No log 55.0 385 2.0133 0.825 0.2839 1.3994 0.825 0.8086 0.1433 0.0560
No log 56.0 392 1.9978 0.835 0.2800 1.4348 0.835 0.8232 0.1415 0.0573
No log 57.0 399 1.9847 0.83 0.2825 1.3907 0.83 0.8153 0.1421 0.0560
No log 58.0 406 1.9892 0.83 0.2832 1.4502 0.83 0.8153 0.1503 0.0566
No log 59.0 413 1.9848 0.83 0.2851 1.4506 0.83 0.8156 0.1462 0.0560
No log 60.0 420 1.9871 0.835 0.2910 1.4527 0.835 0.8191 0.1608 0.0566
No log 61.0 427 1.9914 0.825 0.2932 1.4490 0.825 0.8095 0.1464 0.0587
No log 62.0 434 1.9908 0.825 0.2958 1.4459 0.825 0.8095 0.1493 0.0597
No log 63.0 441 1.9954 0.825 0.3012 1.4480 0.825 0.8095 0.1469 0.0606
No log 64.0 448 2.0111 0.82 0.3050 1.4487 0.82 0.8026 0.1507 0.0619
No log 65.0 455 2.0212 0.82 0.3046 1.4469 0.82 0.8026 0.1604 0.0634
No log 66.0 462 2.0170 0.82 0.3059 1.4443 0.82 0.8040 0.1539 0.0639
No log 67.0 469 2.0170 0.815 0.3056 1.4496 0.815 0.8019 0.1534 0.0643
No log 68.0 476 2.0316 0.82 0.3115 1.4522 0.82 0.8026 0.1606 0.0645
No log 69.0 483 2.0335 0.805 0.3132 1.3831 0.805 0.7855 0.1607 0.0654
No log 70.0 490 2.0362 0.815 0.3106 1.3834 0.815 0.7989 0.1614 0.0655
No log 71.0 497 2.0318 0.815 0.3105 1.3893 0.815 0.7947 0.1541 0.0662
1.3661 72.0 504 2.0434 0.815 0.3135 1.4473 0.815 0.7955 0.1579 0.0653
1.3661 73.0 511 2.0517 0.81 0.3139 1.3838 0.81 0.7917 0.1564 0.0680
1.3661 74.0 518 2.0594 0.82 0.3162 1.3783 0.82 0.7975 0.1626 0.0681
1.3661 75.0 525 2.0628 0.815 0.3210 1.3752 0.815 0.7944 0.1598 0.0706
1.3661 76.0 532 2.0605 0.81 0.3158 1.3711 0.81 0.7886 0.1639 0.0684
1.3661 77.0 539 2.0718 0.815 0.3187 1.3860 0.815 0.7944 0.1710 0.0705
1.3661 78.0 546 2.0749 0.815 0.3168 1.3658 0.815 0.7958 0.1569 0.0713
1.3661 79.0 553 2.0796 0.83 0.3188 1.3016 0.83 0.8147 0.1646 0.0722
1.3661 80.0 560 2.0746 0.81 0.3210 1.3758 0.81 0.7916 0.1580 0.0729
1.3661 81.0 567 2.0819 0.815 0.3194 1.3686 0.815 0.7913 0.1576 0.0722
1.3661 82.0 574 2.0866 0.825 0.3182 1.3627 0.825 0.8085 0.1602 0.0718
1.3661 83.0 581 2.0942 0.815 0.3246 1.3249 0.815 0.8008 0.1591 0.0727
1.3661 84.0 588 2.0938 0.815 0.3246 1.3680 0.815 0.7984 0.1848 0.0727
1.3661 85.0 595 2.0912 0.82 0.3222 1.3662 0.82 0.8012 0.1594 0.0702
1.3661 86.0 602 2.0941 0.82 0.3234 1.3764 0.82 0.8012 0.1576 0.0738
1.3661 87.0 609 2.1037 0.8 0.3304 1.3821 0.8000 0.7821 0.1599 0.0740
1.3661 88.0 616 2.1098 0.805 0.3288 1.3587 0.805 0.7932 0.1678 0.0718
1.3661 89.0 623 2.1119 0.81 0.3276 1.3636 0.81 0.7945 0.1622 0.0728
1.3661 90.0 630 2.1078 0.805 0.3279 1.3641 0.805 0.7914 0.1734 0.0737
1.3661 91.0 637 2.1110 0.8 0.3296 1.3686 0.8000 0.7879 0.1636 0.0744
1.3661 92.0 644 2.1150 0.8 0.3317 1.3685 0.8000 0.7879 0.1730 0.0742
1.3661 93.0 651 2.1146 0.8 0.3303 1.3693 0.8000 0.7881 0.1631 0.0742
1.3661 94.0 658 2.1153 0.805 0.3292 1.3657 0.805 0.7917 0.1676 0.0734
1.3661 95.0 665 2.1188 0.805 0.3298 1.3683 0.805 0.7917 0.1690 0.0735
1.3661 96.0 672 2.1183 0.805 0.3291 1.3691 0.805 0.7914 0.1687 0.0742
1.3661 97.0 679 2.1155 0.81 0.3271 1.3664 0.81 0.7942 0.1599 0.0743
1.3661 98.0 686 2.1183 0.805 0.3285 1.3673 0.805 0.7914 0.1638 0.0740
1.3661 99.0 693 2.1179 0.805 0.3297 1.3686 0.805 0.7917 0.1613 0.0741
1.3661 100.0 700 2.1196 0.805 0.3299 1.3687 0.805 0.7917 0.1606 0.0741

Framework versions

  • Transformers 4.36.0.dev0
  • Pytorch 2.2.0.dev20231112+cu118
  • Datasets 2.14.5
  • Tokenizers 0.14.1
Downloads last month
4
Safetensors
Model size
5.68M params
Tensor type
F32
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for jordyvl/dit-base_tobacco-tiny_tobacco3482_hint

Finetuned
(13)
this model