conditional_detr_finetuned_rsna_2018
This model is a fine-tuned version of microsoft/conditional-detr-resnet-50 on the imagefolder dataset. It achieves the following results on the evaluation set:
- Loss: 1.5677
- Map: 0.1803
- Map 50: 0.4846
- Map 75: 0.0855
- Map Small: 0.0382
- Map Medium: 0.1840
- Map Large: 0.3268
- Mar 1: 0.1569
- Mar 10: 0.4727
- Mar 100: 0.5551
- Mar Small: 0.2600
- Mar Medium: 0.5593
- Mar Large: 0.7209
- Map Pneumonia: 0.1803
- Mar 100 Pneumonia: 0.5551
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: 5e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- num_epochs: 30
Training results
Training Loss | Epoch | Step | Validation Loss | Map | Map 50 | Map 75 | Map Small | Map Medium | Map Large | Mar 1 | Mar 10 | Mar 100 | Mar Small | Mar Medium | Mar Large | Map Pneumonia | Mar 100 Pneumonia |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
No log | 1.0 | 301 | 1.9488 | 0.0513 | 0.1423 | 0.0284 | 0.0377 | 0.0500 | 0.1513 | 0.0992 | 0.2874 | 0.5043 | 0.1433 | 0.5163 | 0.6419 | 0.0513 | 0.5043 |
2.2805 | 2.0 | 602 | 1.8181 | 0.0830 | 0.2514 | 0.0312 | 0.0257 | 0.0860 | 0.2336 | 0.1296 | 0.3520 | 0.5340 | 0.1367 | 0.5524 | 0.6349 | 0.0830 | 0.5340 |
2.2805 | 3.0 | 903 | 1.7791 | 0.1209 | 0.3504 | 0.0546 | 0.0745 | 0.1222 | 0.2082 | 0.1402 | 0.3894 | 0.5315 | 0.1800 | 0.5373 | 0.7209 | 0.1209 | 0.5315 |
1.8321 | 4.0 | 1204 | 1.7326 | 0.1502 | 0.4111 | 0.0696 | 0.0687 | 0.1511 | 0.2655 | 0.1565 | 0.4149 | 0.5234 | 0.1733 | 0.5278 | 0.7256 | 0.1502 | 0.5234 |
1.7202 | 5.0 | 1505 | 1.6750 | 0.1687 | 0.4482 | 0.0909 | 0.0735 | 0.1647 | 0.3322 | 0.1576 | 0.4331 | 0.5540 | 0.2567 | 0.5585 | 0.7186 | 0.1687 | 0.5540 |
1.7202 | 6.0 | 1806 | 1.6909 | 0.1545 | 0.4284 | 0.0581 | 0.0423 | 0.1532 | 0.3184 | 0.1362 | 0.4255 | 0.5414 | 0.2933 | 0.5466 | 0.6651 | 0.1545 | 0.5414 |
1.682 | 7.0 | 2107 | 1.6553 | 0.1773 | 0.4797 | 0.0769 | 0.0545 | 0.1776 | 0.2898 | 0.1582 | 0.4366 | 0.5387 | 0.3667 | 0.5351 | 0.6930 | 0.1773 | 0.5387 |
1.682 | 8.0 | 2408 | 1.6456 | 0.1724 | 0.4565 | 0.0972 | 0.0449 | 0.1717 | 0.3068 | 0.1551 | 0.4437 | 0.5658 | 0.1967 | 0.5766 | 0.7209 | 0.1724 | 0.5658 |
1.6419 | 9.0 | 2709 | 1.6573 | 0.1622 | 0.4276 | 0.0852 | 0.0254 | 0.1647 | 0.2599 | 0.1565 | 0.4387 | 0.5582 | 0.2433 | 0.5646 | 0.7163 | 0.1622 | 0.5582 |
1.6106 | 10.0 | 3010 | 1.6278 | 0.1826 | 0.4852 | 0.0858 | 0.0252 | 0.1842 | 0.3113 | 0.1615 | 0.4441 | 0.5437 | 0.2867 | 0.5488 | 0.6744 | 0.1826 | 0.5437 |
1.6106 | 11.0 | 3311 | 1.6080 | 0.1731 | 0.4664 | 0.0786 | 0.0230 | 0.1781 | 0.2634 | 0.1627 | 0.4609 | 0.5530 | 0.25 | 0.5566 | 0.7302 | 0.1731 | 0.5530 |
1.588 | 12.0 | 3612 | 1.6030 | 0.1839 | 0.4810 | 0.0916 | 0.0460 | 0.1859 | 0.3233 | 0.1617 | 0.4677 | 0.5638 | 0.2233 | 0.5722 | 0.7209 | 0.1839 | 0.5638 |
1.588 | 13.0 | 3913 | 1.6086 | 0.1734 | 0.4531 | 0.0921 | 0.0542 | 0.1737 | 0.3057 | 0.1596 | 0.4621 | 0.5671 | 0.2333 | 0.5759 | 0.7163 | 0.1734 | 0.5671 |
1.5646 | 14.0 | 4214 | 1.5800 | 0.1730 | 0.4562 | 0.0912 | 0.0366 | 0.1747 | 0.3100 | 0.1567 | 0.4644 | 0.5609 | 0.2900 | 0.5639 | 0.7209 | 0.1730 | 0.5609 |
1.5312 | 15.0 | 4515 | 1.5822 | 0.1705 | 0.4566 | 0.0959 | 0.0509 | 0.1730 | 0.2935 | 0.1474 | 0.4654 | 0.5600 | 0.2633 | 0.5646 | 0.7233 | 0.1705 | 0.5600 |
1.5312 | 16.0 | 4816 | 1.5861 | 0.1667 | 0.4606 | 0.0922 | 0.0556 | 0.1673 | 0.3411 | 0.1526 | 0.4731 | 0.5679 | 0.2800 | 0.5707 | 0.7419 | 0.1667 | 0.5679 |
1.5032 | 17.0 | 5117 | 1.5751 | 0.1769 | 0.4898 | 0.0831 | 0.0528 | 0.1766 | 0.3836 | 0.1534 | 0.4615 | 0.5576 | 0.2633 | 0.5627 | 0.7140 | 0.1769 | 0.5576 |
1.5032 | 18.0 | 5418 | 1.5776 | 0.1735 | 0.4671 | 0.0882 | 0.0478 | 0.1745 | 0.3308 | 0.1507 | 0.4675 | 0.5615 | 0.2867 | 0.5641 | 0.7279 | 0.1735 | 0.5615 |
1.4755 | 19.0 | 5719 | 1.5677 | 0.1726 | 0.4612 | 0.0800 | 0.0465 | 0.1745 | 0.3157 | 0.1513 | 0.4729 | 0.5677 | 0.3067 | 0.5707 | 0.7209 | 0.1726 | 0.5677 |
1.4443 | 20.0 | 6020 | 1.5620 | 0.1801 | 0.4676 | 0.0913 | 0.0390 | 0.1828 | 0.3136 | 0.1586 | 0.4640 | 0.5534 | 0.3033 | 0.5534 | 0.7279 | 0.1801 | 0.5534 |
1.4443 | 21.0 | 6321 | 1.5681 | 0.1695 | 0.4654 | 0.0804 | 0.0523 | 0.1718 | 0.3198 | 0.1542 | 0.4702 | 0.5553 | 0.2867 | 0.5578 | 0.7186 | 0.1695 | 0.5553 |
1.4177 | 22.0 | 6622 | 1.5607 | 0.1755 | 0.4630 | 0.0964 | 0.0368 | 0.1784 | 0.3274 | 0.1503 | 0.4803 | 0.5640 | 0.2767 | 0.5698 | 0.7093 | 0.1755 | 0.5640 |
1.4177 | 23.0 | 6923 | 1.5607 | 0.1727 | 0.4731 | 0.0842 | 0.0437 | 0.1761 | 0.3094 | 0.1536 | 0.4708 | 0.5613 | 0.3000 | 0.5634 | 0.7233 | 0.1727 | 0.5613 |
1.3905 | 24.0 | 7224 | 1.5741 | 0.1828 | 0.4882 | 0.0895 | 0.0526 | 0.1836 | 0.3334 | 0.1557 | 0.4729 | 0.5582 | 0.2533 | 0.5629 | 0.7256 | 0.1828 | 0.5582 |
1.3678 | 25.0 | 7525 | 1.5690 | 0.1813 | 0.4763 | 0.0865 | 0.0422 | 0.1849 | 0.3365 | 0.1584 | 0.4683 | 0.5547 | 0.2633 | 0.5583 | 0.7233 | 0.1813 | 0.5547 |
1.3678 | 26.0 | 7826 | 1.5682 | 0.1801 | 0.4813 | 0.0878 | 0.0337 | 0.1854 | 0.3141 | 0.1524 | 0.4733 | 0.5553 | 0.2633 | 0.5598 | 0.7163 | 0.1801 | 0.5553 |
1.3564 | 27.0 | 8127 | 1.5674 | 0.1799 | 0.4849 | 0.0936 | 0.0397 | 0.1841 | 0.3230 | 0.1540 | 0.4743 | 0.5569 | 0.2633 | 0.5615 | 0.7186 | 0.1799 | 0.5569 |
1.3564 | 28.0 | 8428 | 1.5669 | 0.1792 | 0.4826 | 0.0939 | 0.0356 | 0.1832 | 0.3217 | 0.1553 | 0.4733 | 0.5553 | 0.2567 | 0.5602 | 0.7163 | 0.1792 | 0.5553 |
1.3456 | 29.0 | 8729 | 1.5679 | 0.1808 | 0.4848 | 0.0931 | 0.0382 | 0.1844 | 0.3260 | 0.1580 | 0.4733 | 0.5561 | 0.2600 | 0.5607 | 0.7186 | 0.1808 | 0.5561 |
1.3366 | 30.0 | 9030 | 1.5677 | 0.1803 | 0.4846 | 0.0855 | 0.0382 | 0.1840 | 0.3268 | 0.1569 | 0.4727 | 0.5551 | 0.2600 | 0.5593 | 0.7209 | 0.1803 | 0.5551 |
Framework versions
- Transformers 4.43.3
- Pytorch 2.4.0+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1
- Downloads last month
- 3
Model tree for furion-123/conditional_detr_finetuned_rsna_2018
Base model
microsoft/conditional-detr-resnet-50