<!-- 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. -->
cdip-tiny_rvl_cdip-NK1000_kd_NKD_t1.0_g1.5
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: 3.9475
- Accuracy: 0.8255
- Brier Loss: 0.2849
- Nll: 1.5880
- F1 Micro: 0.8255
- F1 Macro: 0.8276
- Ece: 0.1152
- Aurc: 0.0416
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 | 5.0886 | 0.506 | 0.6520 | 2.3481 | 0.506 | 0.4936 | 0.1512 | 0.2571 |
No log | 2.0 | 250 | 4.5869 | 0.639 | 0.5026 | 2.1689 | 0.639 | 0.6339 | 0.1308 | 0.1459 |
No log | 3.0 | 375 | 4.3135 | 0.7065 | 0.4267 | 2.0672 | 0.7065 | 0.7112 | 0.1248 | 0.1057 |
5.1406 | 4.0 | 500 | 4.1859 | 0.717 | 0.3906 | 2.0573 | 0.7170 | 0.7213 | 0.0822 | 0.0921 |
5.1406 | 5.0 | 625 | 3.9770 | 0.7615 | 0.3347 | 1.9365 | 0.7615 | 0.7597 | 0.0587 | 0.0699 |
5.1406 | 6.0 | 750 | 3.9247 | 0.7698 | 0.3189 | 1.8905 | 0.7698 | 0.7710 | 0.0411 | 0.0654 |
5.1406 | 7.0 | 875 | 3.8799 | 0.7728 | 0.3165 | 1.9278 | 0.7728 | 0.7800 | 0.0362 | 0.0644 |
3.778 | 8.0 | 1000 | 3.8415 | 0.786 | 0.3036 | 1.8576 | 0.786 | 0.7927 | 0.0417 | 0.0602 |
3.778 | 9.0 | 1125 | 3.7909 | 0.7977 | 0.2904 | 1.8383 | 0.7977 | 0.7996 | 0.0444 | 0.0547 |
3.778 | 10.0 | 1250 | 3.8216 | 0.794 | 0.2950 | 1.8486 | 0.7940 | 0.7953 | 0.0526 | 0.0534 |
3.778 | 11.0 | 1375 | 3.8084 | 0.797 | 0.2929 | 1.8672 | 0.797 | 0.8004 | 0.0653 | 0.0513 |
3.3945 | 12.0 | 1500 | 3.7547 | 0.8113 | 0.2801 | 1.8312 | 0.8113 | 0.8124 | 0.0548 | 0.0482 |
3.3945 | 13.0 | 1625 | 3.7730 | 0.8137 | 0.2825 | 1.8087 | 0.8137 | 0.8151 | 0.0762 | 0.0482 |
3.3945 | 14.0 | 1750 | 3.8090 | 0.807 | 0.2863 | 1.7529 | 0.807 | 0.8075 | 0.0713 | 0.0504 |
3.3945 | 15.0 | 1875 | 3.7612 | 0.8067 | 0.2886 | 1.7934 | 0.8067 | 0.8113 | 0.0741 | 0.0498 |
3.2666 | 16.0 | 2000 | 3.7760 | 0.809 | 0.2863 | 1.8104 | 0.809 | 0.8116 | 0.0779 | 0.0490 |
3.2666 | 17.0 | 2125 | 3.7504 | 0.8155 | 0.2765 | 1.7438 | 0.8155 | 0.8160 | 0.0798 | 0.0458 |
3.2666 | 18.0 | 2250 | 3.7798 | 0.8085 | 0.2858 | 1.7447 | 0.8085 | 0.8097 | 0.0844 | 0.0462 |
3.2666 | 19.0 | 2375 | 3.7784 | 0.8073 | 0.2876 | 1.7731 | 0.8073 | 0.8112 | 0.0832 | 0.0481 |
3.1995 | 20.0 | 2500 | 3.7772 | 0.8123 | 0.2862 | 1.7110 | 0.8123 | 0.8137 | 0.0887 | 0.0461 |
3.1995 | 21.0 | 2625 | 3.7531 | 0.8155 | 0.2780 | 1.6920 | 0.8155 | 0.8166 | 0.0860 | 0.0435 |
3.1995 | 22.0 | 2750 | 3.7922 | 0.8123 | 0.2850 | 1.7335 | 0.8123 | 0.8162 | 0.0957 | 0.0454 |
3.1995 | 23.0 | 2875 | 3.7857 | 0.8185 | 0.2760 | 1.7026 | 0.8185 | 0.8201 | 0.0870 | 0.0455 |
3.154 | 24.0 | 3000 | 3.7452 | 0.821 | 0.2724 | 1.6936 | 0.821 | 0.8234 | 0.0902 | 0.0425 |
3.154 | 25.0 | 3125 | 3.7485 | 0.8233 | 0.2734 | 1.6908 | 0.8233 | 0.8252 | 0.0889 | 0.0418 |
3.154 | 26.0 | 3250 | 3.7627 | 0.8197 | 0.2754 | 1.6656 | 0.8197 | 0.8212 | 0.0951 | 0.0424 |
3.154 | 27.0 | 3375 | 3.7635 | 0.821 | 0.2743 | 1.6732 | 0.821 | 0.8221 | 0.0955 | 0.0419 |
3.1227 | 28.0 | 3500 | 3.7829 | 0.821 | 0.2765 | 1.6749 | 0.821 | 0.8223 | 0.0980 | 0.0426 |
3.1227 | 29.0 | 3625 | 3.7738 | 0.8207 | 0.2752 | 1.6585 | 0.8207 | 0.8223 | 0.0936 | 0.0417 |
3.1227 | 30.0 | 3750 | 3.7622 | 0.822 | 0.2763 | 1.6672 | 0.822 | 0.8243 | 0.0942 | 0.0421 |
3.1227 | 31.0 | 3875 | 3.7884 | 0.824 | 0.2749 | 1.6566 | 0.824 | 0.8249 | 0.0980 | 0.0416 |
3.0998 | 32.0 | 4000 | 3.7948 | 0.8205 | 0.2780 | 1.6516 | 0.8205 | 0.8225 | 0.1004 | 0.0420 |
3.0998 | 33.0 | 4125 | 3.7831 | 0.8175 | 0.2787 | 1.6503 | 0.8175 | 0.8207 | 0.1022 | 0.0416 |
3.0998 | 34.0 | 4250 | 3.8119 | 0.8223 | 0.2785 | 1.6290 | 0.8223 | 0.8246 | 0.1004 | 0.0421 |
3.0998 | 35.0 | 4375 | 3.8186 | 0.8235 | 0.2798 | 1.6490 | 0.8235 | 0.8263 | 0.1019 | 0.0422 |
3.0845 | 36.0 | 4500 | 3.8304 | 0.8205 | 0.2821 | 1.6117 | 0.8205 | 0.8228 | 0.1062 | 0.0421 |
3.0845 | 37.0 | 4625 | 3.8128 | 0.8267 | 0.2758 | 1.6362 | 0.8267 | 0.8292 | 0.1007 | 0.0409 |
3.0845 | 38.0 | 4750 | 3.8488 | 0.8217 | 0.2812 | 1.6245 | 0.8217 | 0.8236 | 0.1080 | 0.0417 |
3.0845 | 39.0 | 4875 | 3.8459 | 0.826 | 0.2781 | 1.6239 | 0.826 | 0.8281 | 0.1050 | 0.0417 |
3.0726 | 40.0 | 5000 | 3.8527 | 0.8257 | 0.2790 | 1.6083 | 0.8257 | 0.8280 | 0.1078 | 0.0412 |
3.0726 | 41.0 | 5125 | 3.8496 | 0.829 | 0.2777 | 1.6026 | 0.8290 | 0.8304 | 0.1018 | 0.0412 |
3.0726 | 42.0 | 5250 | 3.8656 | 0.826 | 0.2803 | 1.6125 | 0.826 | 0.8283 | 0.1074 | 0.0412 |
3.0726 | 43.0 | 5375 | 3.8860 | 0.8253 | 0.2815 | 1.6029 | 0.8253 | 0.8273 | 0.1102 | 0.0415 |
3.0635 | 44.0 | 5500 | 3.8868 | 0.8225 | 0.2810 | 1.5939 | 0.8225 | 0.8248 | 0.1132 | 0.0414 |
3.0635 | 45.0 | 5625 | 3.9087 | 0.8247 | 0.2825 | 1.5956 | 0.8247 | 0.8268 | 0.1122 | 0.0414 |
3.0635 | 46.0 | 5750 | 3.9273 | 0.8243 | 0.2842 | 1.5863 | 0.8243 | 0.8263 | 0.1150 | 0.0415 |
3.0635 | 47.0 | 5875 | 3.9352 | 0.8247 | 0.2841 | 1.5859 | 0.8247 | 0.8268 | 0.1148 | 0.0416 |
3.0576 | 48.0 | 6000 | 3.9397 | 0.8253 | 0.2843 | 1.5907 | 0.8253 | 0.8274 | 0.1146 | 0.0416 |
3.0576 | 49.0 | 6125 | 3.9444 | 0.8255 | 0.2847 | 1.5886 | 0.8255 | 0.8276 | 0.1147 | 0.0416 |
3.0576 | 50.0 | 6250 | 3.9475 | 0.8255 | 0.2849 | 1.5880 | 0.8255 | 0.8276 | 0.1152 | 0.0416 |
Framework versions
- Transformers 4.26.1
- Pytorch 1.13.1.post200
- Datasets 2.9.0
- Tokenizers 0.13.2