<!-- 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_DsA
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: 0.0287
- Mean Iou: 0.7790
- Mean Accuracy: 0.9499
- Overall Accuracy: 0.9903
- Accuracy Background: nan
- Accuracy Haz: 0.9895
- Accuracy Matrix: 0.9864
- Accuracy Porosity: 0.8091
- Accuracy Carbides: 0.9696
- Accuracy Substrate: 0.9949
- Iou Background: 0.0
- Iou Haz: 0.9826
- Iou Matrix: 0.9705
- Iou Porosity: 0.7840
- Iou Carbides: 0.9495
- Iou Substrate: 0.9874
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: 2
- eval_batch_size: 2
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 15
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 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
0.1399 | 1.0 | 1182 | 0.1095 | 0.6054 | 0.7614 | 0.9642 | nan | 0.9411 | 0.9865 | 0.0 | 0.8913 | 0.9880 | 0.0 | 0.9256 | 0.8643 | 0.0 | 0.8776 | 0.9651 |
0.7874 | 2.0 | 2364 | 0.0563 | 0.6386 | 0.7833 | 0.9809 | nan | 0.9902 | 0.9789 | 0.0105 | 0.9576 | 0.9795 | 0.0 | 0.9633 | 0.9566 | 0.0105 | 0.9288 | 0.9725 |
1.0107 | 3.0 | 3546 | 0.0504 | 0.7051 | 0.8612 | 0.9839 | nan | 0.9844 | 0.9857 | 0.3941 | 0.9533 | 0.9883 | 0.0 | 0.9697 | 0.9590 | 0.3939 | 0.9308 | 0.9775 |
0.8749 | 4.0 | 4728 | 0.0433 | 0.7495 | 0.9150 | 0.9850 | nan | 0.9787 | 0.9877 | 0.6607 | 0.9542 | 0.9939 | 0.0 | 0.9708 | 0.9630 | 0.6484 | 0.9364 | 0.9785 |
0.0469 | 5.0 | 5910 | 0.0477 | 0.7352 | 0.8964 | 0.9831 | nan | 0.9791 | 0.9872 | 0.5815 | 0.9431 | 0.9910 | 0.0 | 0.9680 | 0.9586 | 0.5797 | 0.9285 | 0.9767 |
0.5715 | 6.0 | 7092 | 0.0459 | 0.7520 | 0.9181 | 0.9854 | nan | 0.9821 | 0.9814 | 0.6628 | 0.9729 | 0.9914 | 0.0 | 0.9712 | 0.9658 | 0.6536 | 0.9428 | 0.9785 |
0.5126 | 7.0 | 8274 | 0.0600 | 0.7590 | 0.9282 | 0.9829 | nan | 0.9706 | 0.9838 | 0.7217 | 0.9702 | 0.9944 | 0.0 | 0.9636 | 0.9672 | 0.7060 | 0.9441 | 0.9732 |
0.3617 | 8.0 | 9456 | 0.0367 | 0.7636 | 0.9295 | 0.9879 | nan | 0.9854 | 0.9860 | 0.7157 | 0.9669 | 0.9937 | 0.0 | 0.9772 | 0.9681 | 0.7071 | 0.9457 | 0.9833 |
0.0376 | 9.0 | 10638 | 0.0363 | 0.7763 | 0.9514 | 0.9881 | nan | 0.9847 | 0.9836 | 0.8211 | 0.9731 | 0.9943 | 0.0 | 0.9775 | 0.9693 | 0.7796 | 0.9481 | 0.9836 |
0.2306 | 10.0 | 11820 | 0.0320 | 0.7725 | 0.9448 | 0.9879 | nan | 0.9834 | 0.9859 | 0.7956 | 0.9636 | 0.9953 | 0.0 | 0.9771 | 0.9684 | 0.7613 | 0.9451 | 0.9833 |
0.4667 | 11.0 | 13002 | 0.0335 | 0.7731 | 0.9421 | 0.9889 | nan | 0.9925 | 0.9843 | 0.7716 | 0.9722 | 0.9899 | 0.0 | 0.9795 | 0.9693 | 0.7566 | 0.9482 | 0.9849 |
0.2078 | 12.0 | 14184 | 0.0310 | 0.7758 | 0.9456 | 0.9896 | nan | 0.9892 | 0.9865 | 0.7890 | 0.9696 | 0.9937 | 0.0 | 0.9806 | 0.9703 | 0.7690 | 0.9491 | 0.9858 |
0.0208 | 13.0 | 15366 | 0.0290 | 0.7818 | 0.9574 | 0.9900 | nan | 0.9902 | 0.9861 | 0.8471 | 0.9702 | 0.9936 | 0.0 | 0.9818 | 0.9705 | 0.8025 | 0.9490 | 0.9867 |
0.4266 | 14.0 | 16548 | 0.0290 | 0.7783 | 0.9492 | 0.9903 | nan | 0.9889 | 0.9861 | 0.8050 | 0.9707 | 0.9952 | 0.0 | 0.9828 | 0.9706 | 0.7803 | 0.9488 | 0.9875 |
0.398 | 15.0 | 17730 | 0.0287 | 0.7790 | 0.9499 | 0.9903 | nan | 0.9895 | 0.9864 | 0.8091 | 0.9696 | 0.9949 | 0.0 | 0.9826 | 0.9705 | 0.7840 | 0.9495 | 0.9874 |
Framework versions
- Transformers 4.33.1
- Pytorch 2.0.1+cu118
- Datasets 2.14.5
- Tokenizers 0.13.3