Edit model card

6_e_200-tiny_tobacco3482_kd_CEKD_t5.0_a0.5

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: 0.3932
  • Accuracy: 0.83
  • Brier Loss: 0.2507
  • Nll: 1.3117
  • F1 Micro: 0.83
  • F1 Macro: 0.8164
  • Ece: 0.1915
  • Aurc: 0.0602

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 1.5453 0.21 0.8721 5.2626 0.2100 0.1784 0.2602 0.7492
No log 2.0 50 0.9799 0.515 0.6132 2.7421 0.515 0.4183 0.2927 0.2698
No log 3.0 75 0.7948 0.655 0.4888 2.6042 0.655 0.5597 0.2331 0.1485
No log 4.0 100 0.6422 0.735 0.3906 1.6168 0.735 0.6662 0.2448 0.1095
No log 5.0 125 0.6261 0.785 0.3475 1.1890 0.785 0.7582 0.2616 0.0792
No log 6.0 150 0.6060 0.775 0.3402 1.3317 0.775 0.7269 0.2750 0.0709
No log 7.0 175 0.5659 0.8 0.3459 1.4773 0.8000 0.7741 0.2628 0.0960
No log 8.0 200 0.5339 0.81 0.3038 1.5029 0.81 0.7882 0.2449 0.0699
No log 9.0 225 0.5429 0.805 0.3117 1.4140 0.805 0.7840 0.2242 0.0771
No log 10.0 250 0.5337 0.815 0.3139 1.4630 0.815 0.8167 0.2253 0.0801
No log 11.0 275 0.5257 0.815 0.3084 1.4325 0.815 0.7943 0.2431 0.0823
No log 12.0 300 0.4704 0.81 0.2879 1.3557 0.81 0.7859 0.2139 0.0770
No log 13.0 325 0.4828 0.81 0.2898 1.5643 0.81 0.7767 0.2115 0.0712
No log 14.0 350 0.4579 0.815 0.2733 1.4403 0.815 0.8061 0.1799 0.0609
No log 15.0 375 0.4642 0.815 0.2892 1.4598 0.815 0.8017 0.1973 0.0537
No log 16.0 400 0.4378 0.84 0.2683 1.1278 0.8400 0.8320 0.2135 0.0545
No log 17.0 425 0.4403 0.825 0.2750 1.3817 0.825 0.8024 0.1898 0.0511
No log 18.0 450 0.4211 0.825 0.2646 1.5604 0.825 0.8123 0.1918 0.0538
No log 19.0 475 0.4280 0.82 0.2700 1.5824 0.82 0.8016 0.1814 0.0587
0.4177 20.0 500 0.4223 0.83 0.2649 1.8744 0.83 0.8152 0.1885 0.0681
0.4177 21.0 525 0.4219 0.835 0.2682 1.6053 0.835 0.8252 0.1932 0.0600
0.4177 22.0 550 0.3935 0.835 0.2534 1.4134 0.835 0.8229 0.2049 0.0680
0.4177 23.0 575 0.4231 0.82 0.2651 1.7267 0.82 0.8028 0.1943 0.0636
0.4177 24.0 600 0.4135 0.845 0.2576 1.8708 0.845 0.8304 0.1872 0.0559
0.4177 25.0 625 0.4027 0.835 0.2526 1.5970 0.835 0.8187 0.1886 0.0645
0.4177 26.0 650 0.4000 0.835 0.2513 1.7233 0.835 0.8151 0.1998 0.0613
0.4177 27.0 675 0.3956 0.83 0.2478 1.5716 0.83 0.8143 0.1871 0.0533
0.4177 28.0 700 0.3952 0.835 0.2493 1.5638 0.835 0.8153 0.1942 0.0593
0.4177 29.0 725 0.3965 0.83 0.2509 1.5069 0.83 0.8150 0.1811 0.0594
0.4177 30.0 750 0.3935 0.835 0.2486 1.5657 0.835 0.8168 0.1948 0.0573
0.4177 31.0 775 0.3951 0.83 0.2498 1.5061 0.83 0.8146 0.1886 0.0591
0.4177 32.0 800 0.3940 0.835 0.2506 1.5054 0.835 0.8203 0.1991 0.0598
0.4177 33.0 825 0.3935 0.84 0.2493 1.5025 0.8400 0.8230 0.1932 0.0590
0.4177 34.0 850 0.3928 0.84 0.2485 1.5679 0.8400 0.8227 0.1981 0.0570
0.4177 35.0 875 0.3942 0.835 0.2497 1.5670 0.835 0.8177 0.1940 0.0592
0.4177 36.0 900 0.3935 0.835 0.2491 1.5120 0.835 0.8203 0.1931 0.0596
0.4177 37.0 925 0.3932 0.835 0.2496 1.5715 0.835 0.8203 0.1989 0.0597
0.4177 38.0 950 0.3924 0.835 0.2491 1.5091 0.835 0.8203 0.1973 0.0606
0.4177 39.0 975 0.3936 0.835 0.2500 1.5036 0.835 0.8203 0.1954 0.0602
0.0597 40.0 1000 0.3936 0.835 0.2497 1.4602 0.835 0.8203 0.2053 0.0597
0.0597 41.0 1025 0.3936 0.835 0.2505 1.5040 0.835 0.8203 0.2026 0.0607
0.0597 42.0 1050 0.3931 0.83 0.2500 1.4565 0.83 0.8164 0.1961 0.0590
0.0597 43.0 1075 0.3931 0.835 0.2497 1.5208 0.835 0.8203 0.1972 0.0591
0.0597 44.0 1100 0.3932 0.835 0.2503 1.5040 0.835 0.8203 0.2030 0.0606
0.0597 45.0 1125 0.3930 0.835 0.2502 1.4555 0.835 0.8203 0.1992 0.0604
0.0597 46.0 1150 0.3927 0.835 0.2500 1.4553 0.835 0.8203 0.1960 0.0616
0.0597 47.0 1175 0.3928 0.835 0.2501 1.3970 0.835 0.8203 0.1965 0.0610
0.0597 48.0 1200 0.3930 0.835 0.2498 1.3967 0.835 0.8203 0.1989 0.0599
0.0597 49.0 1225 0.3931 0.835 0.2502 1.4578 0.835 0.8203 0.1963 0.0606
0.0597 50.0 1250 0.3932 0.835 0.2504 1.4475 0.835 0.8203 0.1996 0.0604
0.0597 51.0 1275 0.3928 0.835 0.2500 1.3382 0.835 0.8203 0.2002 0.0609
0.0597 52.0 1300 0.3933 0.83 0.2502 1.4424 0.83 0.8164 0.1991 0.0597
0.0597 53.0 1325 0.3933 0.83 0.2502 1.3390 0.83 0.8164 0.1965 0.0604
0.0597 54.0 1350 0.3929 0.83 0.2502 1.3351 0.83 0.8164 0.1914 0.0608
0.0597 55.0 1375 0.3932 0.83 0.2503 1.3422 0.83 0.8164 0.1969 0.0608
0.0597 56.0 1400 0.3934 0.83 0.2506 1.3369 0.83 0.8164 0.1950 0.0599
0.0597 57.0 1425 0.3930 0.83 0.2502 1.3829 0.83 0.8164 0.1966 0.0603
0.0597 58.0 1450 0.3930 0.835 0.2503 1.3219 0.835 0.8203 0.1907 0.0607
0.0597 59.0 1475 0.3930 0.83 0.2504 1.3268 0.83 0.8164 0.1919 0.0599
0.0574 60.0 1500 0.3933 0.835 0.2505 1.3242 0.835 0.8203 0.1913 0.0601
0.0574 61.0 1525 0.3930 0.83 0.2504 1.3205 0.83 0.8164 0.1943 0.0607
0.0574 62.0 1550 0.3930 0.83 0.2504 1.3189 0.83 0.8164 0.1947 0.0608
0.0574 63.0 1575 0.3931 0.83 0.2504 1.3197 0.83 0.8164 0.1917 0.0600
0.0574 64.0 1600 0.3931 0.835 0.2505 1.3200 0.835 0.8203 0.1904 0.0597
0.0574 65.0 1625 0.3932 0.83 0.2505 1.3175 0.83 0.8164 0.1915 0.0601
0.0574 66.0 1650 0.3931 0.83 0.2506 1.3200 0.83 0.8164 0.1917 0.0608
0.0574 67.0 1675 0.3929 0.83 0.2503 1.3188 0.83 0.8164 0.1940 0.0598
0.0574 68.0 1700 0.3931 0.83 0.2505 1.3160 0.83 0.8164 0.1913 0.0599
0.0574 69.0 1725 0.3931 0.83 0.2505 1.3161 0.83 0.8164 0.1941 0.0598
0.0574 70.0 1750 0.3932 0.83 0.2506 1.3171 0.83 0.8164 0.1961 0.0608
0.0574 71.0 1775 0.3930 0.83 0.2506 1.3161 0.83 0.8164 0.1913 0.0602
0.0574 72.0 1800 0.3929 0.83 0.2505 1.3155 0.83 0.8164 0.1960 0.0603
0.0574 73.0 1825 0.3930 0.83 0.2506 1.3152 0.83 0.8164 0.1941 0.0601
0.0574 74.0 1850 0.3930 0.83 0.2506 1.3167 0.83 0.8164 0.1940 0.0602
0.0574 75.0 1875 0.3933 0.83 0.2507 1.3148 0.83 0.8164 0.1918 0.0600
0.0574 76.0 1900 0.3930 0.83 0.2505 1.3146 0.83 0.8164 0.1914 0.0602
0.0574 77.0 1925 0.3930 0.83 0.2505 1.3147 0.83 0.8164 0.1914 0.0598
0.0574 78.0 1950 0.3931 0.83 0.2506 1.3134 0.83 0.8164 0.1942 0.0601
0.0574 79.0 1975 0.3931 0.83 0.2505 1.3137 0.83 0.8164 0.1916 0.0598
0.0573 80.0 2000 0.3931 0.83 0.2506 1.3136 0.83 0.8164 0.1915 0.0601
0.0573 81.0 2025 0.3932 0.83 0.2506 1.3132 0.83 0.8164 0.1915 0.0607
0.0573 82.0 2050 0.3933 0.83 0.2507 1.3142 0.83 0.8164 0.1943 0.0603
0.0573 83.0 2075 0.3933 0.83 0.2507 1.3135 0.83 0.8164 0.1916 0.0603
0.0573 84.0 2100 0.3931 0.83 0.2506 1.3124 0.83 0.8164 0.1914 0.0601
0.0573 85.0 2125 0.3931 0.83 0.2507 1.3128 0.83 0.8164 0.1915 0.0602
0.0573 86.0 2150 0.3931 0.83 0.2506 1.3128 0.83 0.8164 0.1916 0.0602
0.0573 87.0 2175 0.3932 0.83 0.2507 1.3130 0.83 0.8164 0.1943 0.0602
0.0573 88.0 2200 0.3932 0.83 0.2507 1.3123 0.83 0.8164 0.1915 0.0602
0.0573 89.0 2225 0.3932 0.83 0.2507 1.3123 0.83 0.8164 0.1915 0.0599
0.0573 90.0 2250 0.3931 0.83 0.2507 1.3119 0.83 0.8164 0.1915 0.0602
0.0573 91.0 2275 0.3932 0.83 0.2507 1.3121 0.83 0.8164 0.1915 0.0602
0.0573 92.0 2300 0.3932 0.83 0.2507 1.3117 0.83 0.8164 0.1915 0.0601
0.0573 93.0 2325 0.3931 0.83 0.2507 1.3117 0.83 0.8164 0.1915 0.0602
0.0573 94.0 2350 0.3932 0.83 0.2507 1.3120 0.83 0.8164 0.1915 0.0602
0.0573 95.0 2375 0.3932 0.83 0.2507 1.3120 0.83 0.8164 0.1916 0.0602
0.0573 96.0 2400 0.3932 0.83 0.2507 1.3119 0.83 0.8164 0.1915 0.0602
0.0573 97.0 2425 0.3932 0.83 0.2507 1.3118 0.83 0.8164 0.1915 0.0601
0.0573 98.0 2450 0.3932 0.83 0.2507 1.3117 0.83 0.8164 0.1915 0.0602
0.0573 99.0 2475 0.3932 0.83 0.2507 1.3118 0.83 0.8164 0.1915 0.0602
0.0573 100.0 2500 0.3932 0.83 0.2507 1.3117 0.83 0.8164 0.1915 0.0602

Framework versions

  • Transformers 4.30.2
  • Pytorch 1.13.1
  • Datasets 2.13.1
  • Tokenizers 0.13.3
Downloads last month
0
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.