Edit model card

segformer-b5-finetuned-magic-cards-230117-2

This model is a fine-tuned version of nvidia/mit-b5 on the andrewljohnson/magic_cards dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0491
  • Mean Iou: 0.6649
  • Mean Accuracy: 0.9974
  • Overall Accuracy: 0.9972
  • Accuracy Unlabeled: nan
  • Accuracy Front: 0.9990
  • Accuracy Back: 0.9957
  • Iou Unlabeled: 0.0
  • Iou Front: 0.9990
  • Iou Back: 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: 1
  • eval_batch_size: 1
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 5

Training results

Training Loss Epoch Step Validation Loss Mean Iou Mean Accuracy Overall Accuracy Accuracy Unlabeled Accuracy Front Accuracy Back Iou Unlabeled Iou Front Iou Back
0.5968 0.33 20 0.4422 0.6366 0.9701 0.9690 nan 0.9812 0.9590 0.0 0.9507 0.9590
0.8955 0.66 40 0.2353 0.6496 0.9819 0.9807 nan 0.9944 0.9695 0.0 0.9792 0.9695
0.1269 0.98 60 0.1739 0.6566 0.9922 0.9916 nan 0.9979 0.9866 0.0 0.9832 0.9866
0.7629 1.31 80 0.1664 0.6561 0.9915 0.9909 nan 0.9975 0.9856 0.0 0.9826 0.9856
0.106 1.64 100 0.1005 0.6641 0.9968 0.9967 nan 0.9978 0.9959 0.0 0.9966 0.9959
0.3278 1.97 120 0.0577 0.6632 0.9948 0.9947 nan 0.9963 0.9934 0.0 0.9963 0.9934
0.061 2.3 140 0.0655 0.6642 0.9963 0.9962 nan 0.9972 0.9953 0.0 0.9972 0.9953
0.0766 2.62 160 0.0470 0.6635 0.9953 0.9954 nan 0.9940 0.9966 0.0 0.9940 0.9966
0.0664 2.95 180 0.0436 0.6617 0.9926 0.9931 nan 0.9877 0.9975 0.0 0.9877 0.9975
0.0655 3.28 200 0.0632 0.6649 0.9973 0.9971 nan 0.9994 0.9953 0.0 0.9994 0.9953
0.0356 3.61 220 0.0755 0.6661 0.9991 0.9991 nan 0.9992 0.9991 0.0 0.9992 0.9991
0.0516 3.93 240 0.0470 0.6643 0.9965 0.9963 nan 0.9987 0.9943 0.0 0.9987 0.9943
0.0517 4.26 260 0.0481 0.6645 0.9967 0.9965 nan 0.9989 0.9945 0.0 0.9989 0.9945
0.1886 4.59 280 0.0823 0.6659 0.9988 0.9987 nan 0.9999 0.9977 0.0 0.9999 0.9977
0.0453 4.92 300 0.0491 0.6649 0.9974 0.9972 nan 0.9990 0.9957 0.0 0.9990 0.9957

Framework versions

  • Transformers 4.25.1
  • Pytorch 1.12.1
  • Datasets 2.8.0
  • Tokenizers 0.13.0.dev0
Downloads last month
19