<!-- 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. -->
segformer-b0_DsA1
This model is a fine-tuned version of nvidia/mit-b0 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.0115
- Mean Iou: 0.8020
- Mean Accuracy: 0.9814
- Overall Accuracy: 0.9942
- Accuracy Background: nan
- Accuracy Haz: 0.9968
- Accuracy Matrix: 0.9873
- Accuracy Porosity: 0.9517
- Accuracy Carbides: 0.9730
- Accuracy Substrate: 0.9980
- Iou Background: 0.0
- Iou Haz: 0.9940
- Iou Matrix: 0.9748
- Iou Porosity: 0.8955
- Iou Carbides: 0.9519
- Iou Substrate: 0.9957
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: 6e-05
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 30
Training results
Training Loss | Epoch | Step | Validation Loss | Mean Iou | Mean Accuracy | Overall Accuracy | Accuracy Background | Accuracy Haz | Accuracy Matrix | Accuracy Porosity | Accuracy Carbides | Accuracy Substrate | Iou Background | Iou Haz | Iou Matrix | Iou Porosity | Iou Carbides | Iou Substrate |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
0.4915 | 1.0 | 390 | 0.0877 | 0.6229 | 0.7723 | 0.9723 | nan | 0.9770 | 0.9720 | 0.0 | 0.9346 | 0.9779 | 0.0 | 0.9501 | 0.9367 | 0.0 | 0.8867 | 0.9638 |
0.3601 | 2.0 | 780 | 0.0608 | 0.6259 | 0.7710 | 0.9744 | nan | 0.9881 | 0.9849 | 0.0 | 0.9085 | 0.9734 | 0.0 | 0.9556 | 0.9423 | 0.0 | 0.8919 | 0.9654 |
1.9097 | 3.0 | 1170 | 0.0419 | 0.6385 | 0.7837 | 0.9826 | nan | 0.9820 | 0.9784 | 0.0 | 0.9682 | 0.9901 | 0.0 | 0.9695 | 0.9610 | 0.0 | 0.9236 | 0.9767 |
2.401 | 4.0 | 1560 | 0.0658 | 0.6718 | 0.8261 | 0.9744 | nan | 0.9948 | 0.9852 | 0.2337 | 0.9574 | 0.9594 | 0.0 | 0.9465 | 0.9642 | 0.2333 | 0.9311 | 0.9556 |
2.4115 | 5.0 | 1950 | 0.0253 | 0.7479 | 0.9119 | 0.9880 | nan | 0.9925 | 0.9850 | 0.6293 | 0.9618 | 0.9909 | 0.0 | 0.9807 | 0.9669 | 0.6154 | 0.9389 | 0.9852 |
1.0366 | 6.0 | 2340 | 0.0523 | 0.7747 | 0.9475 | 0.9825 | nan | 0.9681 | 0.9832 | 0.8235 | 0.9653 | 0.9972 | 0.0 | 0.9633 | 0.9670 | 0.8049 | 0.9408 | 0.9722 |
1.4287 | 7.0 | 2730 | 0.0217 | 0.7834 | 0.9626 | 0.9888 | nan | 0.9944 | 0.9879 | 0.8795 | 0.9619 | 0.9893 | 0.0 | 0.9811 | 0.9701 | 0.8212 | 0.9426 | 0.9852 |
1.4637 | 8.0 | 3120 | 0.0212 | 0.7908 | 0.9692 | 0.9893 | nan | 0.9960 | 0.9822 | 0.9038 | 0.9752 | 0.9890 | 0.0 | 0.9822 | 0.9705 | 0.8611 | 0.9449 | 0.9861 |
1.2401 | 9.0 | 3510 | 0.0204 | 0.7905 | 0.9679 | 0.9902 | nan | 0.9921 | 0.9864 | 0.8999 | 0.9674 | 0.9939 | 0.0 | 0.9844 | 0.9712 | 0.8538 | 0.9453 | 0.9880 |
1.2133 | 10.0 | 3900 | 0.0181 | 0.7915 | 0.9714 | 0.9907 | nan | 0.9935 | 0.9860 | 0.9164 | 0.9673 | 0.9940 | 0.0 | 0.9858 | 0.9709 | 0.8578 | 0.9451 | 0.9892 |
0.0667 | 11.0 | 4290 | 0.0172 | 0.7945 | 0.9707 | 0.9913 | nan | 0.9917 | 0.9839 | 0.9064 | 0.9751 | 0.9965 | 0.0 | 0.9869 | 0.9722 | 0.8713 | 0.9469 | 0.9900 |
0.0292 | 12.0 | 4680 | 0.0168 | 0.7959 | 0.9723 | 0.9916 | nan | 0.9900 | 0.9867 | 0.9174 | 0.9687 | 0.9985 | 0.0 | 0.9876 | 0.9722 | 0.8774 | 0.9473 | 0.9906 |
1.0385 | 13.0 | 5070 | 0.0169 | 0.7955 | 0.9757 | 0.9914 | nan | 0.9981 | 0.9837 | 0.9300 | 0.9750 | 0.9916 | 0.0 | 0.9874 | 0.9721 | 0.8753 | 0.9476 | 0.9903 |
1.8648 | 14.0 | 5460 | 0.0231 | 0.7953 | 0.9733 | 0.9914 | nan | 0.9893 | 0.9856 | 0.9211 | 0.9722 | 0.9984 | 0.0 | 0.9871 | 0.9721 | 0.8744 | 0.9480 | 0.9904 |
0.1106 | 15.0 | 5850 | 0.0141 | 0.7972 | 0.9744 | 0.9927 | nan | 0.9965 | 0.9875 | 0.9228 | 0.9696 | 0.9954 | 0.0 | 0.9906 | 0.9733 | 0.8776 | 0.9488 | 0.9930 |
0.1963 | 16.0 | 6240 | 0.0131 | 0.7985 | 0.9756 | 0.9933 | nan | 0.9962 | 0.9876 | 0.9270 | 0.9703 | 0.9968 | 0.0 | 0.9921 | 0.9735 | 0.8822 | 0.9490 | 0.9944 |
0.0299 | 17.0 | 6630 | 0.0146 | 0.7984 | 0.9795 | 0.9930 | nan | 0.9942 | 0.9869 | 0.9467 | 0.9720 | 0.9976 | 0.0 | 0.9910 | 0.9738 | 0.8824 | 0.9500 | 0.9934 |
0.0464 | 18.0 | 7020 | 0.0126 | 0.7997 | 0.9796 | 0.9935 | nan | 0.9952 | 0.9859 | 0.9442 | 0.9749 | 0.9980 | 0.0 | 0.9924 | 0.9740 | 0.8871 | 0.9502 | 0.9945 |
0.0197 | 19.0 | 7410 | 0.0132 | 0.8002 | 0.9792 | 0.9931 | nan | 0.9981 | 0.9837 | 0.9408 | 0.9783 | 0.9950 | 0.0 | 0.9915 | 0.9735 | 0.8921 | 0.9502 | 0.9937 |
0.0691 | 20.0 | 7800 | 0.0131 | 0.8007 | 0.9793 | 0.9934 | nan | 0.9974 | 0.9832 | 0.9398 | 0.9798 | 0.9961 | 0.0 | 0.9923 | 0.9735 | 0.8939 | 0.9499 | 0.9944 |
0.0384 | 21.0 | 8190 | 0.0121 | 0.8004 | 0.9810 | 0.9937 | nan | 0.9970 | 0.9855 | 0.9504 | 0.9751 | 0.9971 | 0.0 | 0.9931 | 0.9740 | 0.8895 | 0.9510 | 0.9950 |
0.0125 | 22.0 | 8580 | 0.0115 | 0.8019 | 0.9779 | 0.9941 | nan | 0.9967 | 0.9862 | 0.9330 | 0.9757 | 0.9979 | 0.0 | 0.9938 | 0.9746 | 0.8962 | 0.9514 | 0.9955 |
0.8233 | 23.0 | 8970 | 0.0119 | 0.8012 | 0.9788 | 0.9938 | nan | 0.9972 | 0.9850 | 0.9372 | 0.9774 | 0.9969 | 0.0 | 0.9933 | 0.9741 | 0.8937 | 0.9511 | 0.9951 |
2.932 | 24.0 | 9360 | 0.0132 | 0.8008 | 0.9762 | 0.9938 | nan | 0.9968 | 0.9882 | 0.9270 | 0.9722 | 0.9970 | 0.0 | 0.9928 | 0.9749 | 0.8912 | 0.9513 | 0.9947 |
0.056 | 25.0 | 9750 | 0.0122 | 0.8014 | 0.9783 | 0.9939 | nan | 0.9959 | 0.9864 | 0.9358 | 0.9755 | 0.9981 | 0.0 | 0.9932 | 0.9747 | 0.8942 | 0.9514 | 0.9950 |
0.0626 | 26.0 | 10140 | 0.0116 | 0.8016 | 0.9799 | 0.9940 | nan | 0.9955 | 0.9857 | 0.9427 | 0.9767 | 0.9986 | 0.0 | 0.9937 | 0.9746 | 0.8945 | 0.9517 | 0.9955 |
0.0146 | 27.0 | 10530 | 0.0112 | 0.8019 | 0.9782 | 0.9943 | nan | 0.9966 | 0.9865 | 0.9338 | 0.9757 | 0.9983 | 0.0 | 0.9942 | 0.9749 | 0.8950 | 0.9518 | 0.9958 |
0.0879 | 28.0 | 10920 | 0.0117 | 0.8016 | 0.9795 | 0.9940 | nan | 0.9969 | 0.9886 | 0.9442 | 0.9699 | 0.9976 | 0.0 | 0.9937 | 0.9746 | 0.8947 | 0.9510 | 0.9955 |
0.0165 | 29.0 | 11310 | 0.0114 | 0.8020 | 0.9787 | 0.9942 | nan | 0.9966 | 0.9883 | 0.9385 | 0.9719 | 0.9981 | 0.0 | 0.9940 | 0.9750 | 0.8958 | 0.9517 | 0.9957 |
0.6593 | 30.0 | 11700 | 0.0115 | 0.8020 | 0.9814 | 0.9942 | nan | 0.9968 | 0.9873 | 0.9517 | 0.9730 | 0.9980 | 0.0 | 0.9940 | 0.9748 | 0.8955 | 0.9519 | 0.9957 |
Framework versions
- Transformers 4.33.1
- Pytorch 2.0.1+cu118
- Datasets 2.14.5
- Tokenizers 0.13.3