---
library_name: transformers
tags: []
---

## Original result
```
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.011
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.011
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.011
```

## After training result
```
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.008
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.026
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.001
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.008
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.040
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.049
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.056
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.056
```

## Config
- dataset: NIH
- original model: hustvl/yolos-tiny
- lr: 0.0001
- dropout_rate: 0.1
- weight_decay: 10.0
- max_epochs: 30
- train samples: 885

## Logging
### Training process
```
{'validation_loss': tensor(6.4845, device='cuda:0'), 'validation_loss_ce': tensor(2.0703, device='cuda:0'), 'validation_loss_bbox': tensor(0.5406, device='cuda:0'), 'validation_loss_giou': tensor(0.8556, device='cuda:0'), 'validation_cardinality_error': tensor(79.9062, device='cuda:0')}
{'training_loss': tensor(2.7016, device='cuda:0'), 'train_loss_ce': tensor(0.4314, device='cuda:0'), 'train_loss_bbox': tensor(0.2228, device='cuda:0'), 'train_loss_giou': tensor(0.5781, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.3099, device='cuda:0'), 'validation_loss_ce': tensor(0.4543, device='cuda:0'), 'validation_loss_bbox': tensor(0.1610, device='cuda:0'), 'validation_loss_giou': tensor(0.5252, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.2841, device='cuda:0'), 'train_loss_ce': tensor(0.4562, device='cuda:0'), 'train_loss_bbox': tensor(0.1660, device='cuda:0'), 'train_loss_giou': tensor(0.4990, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.3262, device='cuda:0'), 'validation_loss_ce': tensor(0.4318, device='cuda:0'), 'validation_loss_bbox': tensor(0.1678, device='cuda:0'), 'validation_loss_giou': tensor(0.5276, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.0995, device='cuda:0'), 'train_loss_ce': tensor(0.4676, device='cuda:0'), 'train_loss_bbox': tensor(0.1685, device='cuda:0'), 'train_loss_giou': tensor(0.3948, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.2856, device='cuda:0'), 'validation_loss_ce': tensor(0.4396, device='cuda:0'), 'validation_loss_bbox': tensor(0.1619, device='cuda:0'), 'validation_loss_giou': tensor(0.5182, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(1.8772, device='cuda:0'), 'train_loss_ce': tensor(0.4828, device='cuda:0'), 'train_loss_bbox': tensor(0.1217, device='cuda:0'), 'train_loss_giou': tensor(0.3929, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.2444, device='cuda:0'), 'validation_loss_ce': tensor(0.4270, device='cuda:0'), 'validation_loss_bbox': tensor(0.1613, device='cuda:0'), 'validation_loss_giou': tensor(0.5056, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.4697, device='cuda:0'), 'train_loss_ce': tensor(0.4485, device='cuda:0'), 'train_loss_bbox': tensor(0.1694, device='cuda:0'), 'train_loss_giou': tensor(0.5871, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.3423, device='cuda:0'), 'validation_loss_ce': tensor(0.4061, device='cuda:0'), 'validation_loss_bbox': tensor(0.1712, device='cuda:0'), 'validation_loss_giou': tensor(0.5401, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(3.2742, device='cuda:0'), 'train_loss_ce': tensor(0.3894, device='cuda:0'), 'train_loss_bbox': tensor(0.2832, device='cuda:0'), 'train_loss_giou': tensor(0.7344, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.8731, device='cuda:0'), 'validation_loss_ce': tensor(0.4232, device='cuda:0'), 'validation_loss_bbox': tensor(0.2260, device='cuda:0'), 'validation_loss_giou': tensor(0.6598, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.7366, device='cuda:0'), 'train_loss_ce': tensor(0.4021, device='cuda:0'), 'train_loss_bbox': tensor(0.1880, device='cuda:0'), 'train_loss_giou': tensor(0.6971, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.7447, device='cuda:0'), 'validation_loss_ce': tensor(0.4046, device='cuda:0'), 'validation_loss_bbox': tensor(0.2111, device='cuda:0'), 'validation_loss_giou': tensor(0.6423, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(3.5912, device='cuda:0'), 'train_loss_ce': tensor(0.4474, device='cuda:0'), 'train_loss_bbox': tensor(0.2692, device='cuda:0'), 'train_loss_giou': tensor(0.8990, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.7000, device='cuda:0'), 'validation_loss_ce': tensor(0.3900, device='cuda:0'), 'validation_loss_bbox': tensor(0.2162, device='cuda:0'), 'validation_loss_giou': tensor(0.6146, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(3.1424, device='cuda:0'), 'train_loss_ce': tensor(0.4654, device='cuda:0'), 'train_loss_bbox': tensor(0.2374, device='cuda:0'), 'train_loss_giou': tensor(0.7449, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.6695, device='cuda:0'), 'validation_loss_ce': tensor(0.4214, device='cuda:0'), 'validation_loss_bbox': tensor(0.2038, device='cuda:0'), 'validation_loss_giou': tensor(0.6146, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(3.3901, device='cuda:0'), 'train_loss_ce': tensor(0.3745, device='cuda:0'), 'train_loss_bbox': tensor(0.3009, device='cuda:0'), 'train_loss_giou': tensor(0.7555, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.8418, device='cuda:0'), 'validation_loss_ce': tensor(0.4005, device='cuda:0'), 'validation_loss_bbox': tensor(0.2356, device='cuda:0'), 'validation_loss_giou': tensor(0.6315, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.1228, device='cuda:0'), 'train_loss_ce': tensor(0.4134, device='cuda:0'), 'train_loss_bbox': tensor(0.1691, device='cuda:0'), 'train_loss_giou': tensor(0.4320, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.8125, device='cuda:0'), 'validation_loss_ce': tensor(0.4067, device='cuda:0'), 'validation_loss_bbox': tensor(0.2164, device='cuda:0'), 'validation_loss_giou': tensor(0.6619, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.8465, device='cuda:0'), 'train_loss_ce': tensor(0.4752, device='cuda:0'), 'train_loss_bbox': tensor(0.2065, device='cuda:0'), 'train_loss_giou': tensor(0.6694, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.5769, device='cuda:0'), 'validation_loss_ce': tensor(0.4146, device='cuda:0'), 'validation_loss_bbox': tensor(0.1986, device='cuda:0'), 'validation_loss_giou': tensor(0.5845, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(3.0633, device='cuda:0'), 'train_loss_ce': tensor(0.4504, device='cuda:0'), 'train_loss_bbox': tensor(0.1999, device='cuda:0'), 'train_loss_giou': tensor(0.8068, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.7978, device='cuda:0'), 'validation_loss_ce': tensor(0.4000, device='cuda:0'), 'validation_loss_bbox': tensor(0.2214, device='cuda:0'), 'validation_loss_giou': tensor(0.6452, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.3091, device='cuda:0'), 'train_loss_ce': tensor(0.4060, device='cuda:0'), 'train_loss_bbox': tensor(0.1832, device='cuda:0'), 'train_loss_giou': tensor(0.4936, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.6313, device='cuda:0'), 'validation_loss_ce': tensor(0.4259, device='cuda:0'), 'validation_loss_bbox': tensor(0.2090, device='cuda:0'), 'validation_loss_giou': tensor(0.5803, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.8676, device='cuda:0'), 'train_loss_ce': tensor(0.4144, device='cuda:0'), 'train_loss_bbox': tensor(0.2126, device='cuda:0'), 'train_loss_giou': tensor(0.6952, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.6389, device='cuda:0'), 'validation_loss_ce': tensor(0.4249, device='cuda:0'), 'validation_loss_bbox': tensor(0.2028, device='cuda:0'), 'validation_loss_giou': tensor(0.6000, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.8933, device='cuda:0'), 'train_loss_ce': tensor(0.4095, device='cuda:0'), 'train_loss_bbox': tensor(0.2215, device='cuda:0'), 'train_loss_giou': tensor(0.6881, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.7148, device='cuda:0'), 'validation_loss_ce': tensor(0.4285, device='cuda:0'), 'validation_loss_bbox': tensor(0.2068, device='cuda:0'), 'validation_loss_giou': tensor(0.6262, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.8105, device='cuda:0'), 'train_loss_ce': tensor(0.4766, device='cuda:0'), 'train_loss_bbox': tensor(0.2018, device='cuda:0'), 'train_loss_giou': tensor(0.6625, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.6955, device='cuda:0'), 'validation_loss_ce': tensor(0.4442, device='cuda:0'), 'validation_loss_bbox': tensor(0.2032, device='cuda:0'), 'validation_loss_giou': tensor(0.6176, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.9315, device='cuda:0'), 'train_loss_ce': tensor(0.5202, device='cuda:0'), 'train_loss_bbox': tensor(0.2002, device='cuda:0'), 'train_loss_giou': tensor(0.7052, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.9419, device='cuda:0'), 'validation_loss_ce': tensor(0.4325, device='cuda:0'), 'validation_loss_bbox': tensor(0.2266, device='cuda:0'), 'validation_loss_giou': tensor(0.6881, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.4729, device='cuda:0'), 'train_loss_ce': tensor(0.4755, device='cuda:0'), 'train_loss_bbox': tensor(0.1909, device='cuda:0'), 'train_loss_giou': tensor(0.5214, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.7052, device='cuda:0'), 'validation_loss_ce': tensor(0.4322, device='cuda:0'), 'validation_loss_bbox': tensor(0.2143, device='cuda:0'), 'validation_loss_giou': tensor(0.6007, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.7326, device='cuda:0'), 'train_loss_ce': tensor(0.3718, device='cuda:0'), 'train_loss_bbox': tensor(0.2027, device='cuda:0'), 'train_loss_giou': tensor(0.6737, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.7806, device='cuda:0'), 'validation_loss_ce': tensor(0.4508, device='cuda:0'), 'validation_loss_bbox': tensor(0.2179, device='cuda:0'), 'validation_loss_giou': tensor(0.6201, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.5336, device='cuda:0'), 'train_loss_ce': tensor(0.4251, device='cuda:0'), 'train_loss_bbox': tensor(0.2083, device='cuda:0'), 'train_loss_giou': tensor(0.5336, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.9040, device='cuda:0'), 'validation_loss_ce': tensor(0.4602, device='cuda:0'), 'validation_loss_bbox': tensor(0.2258, device='cuda:0'), 'validation_loss_giou': tensor(0.6575, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.5256, device='cuda:0'), 'train_loss_ce': tensor(0.4589, device='cuda:0'), 'train_loss_bbox': tensor(0.2075, device='cuda:0'), 'train_loss_giou': tensor(0.5147, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.6811, device='cuda:0'), 'validation_loss_ce': tensor(0.4428, device='cuda:0'), 'validation_loss_bbox': tensor(0.2018, device='cuda:0'), 'validation_loss_giou': tensor(0.6145, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.0975, device='cuda:0'), 'train_loss_ce': tensor(0.4765, device='cuda:0'), 'train_loss_bbox': tensor(0.1326, device='cuda:0'), 'train_loss_giou': tensor(0.4790, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.7996, device='cuda:0'), 'validation_loss_ce': tensor(0.4371, device='cuda:0'), 'validation_loss_bbox': tensor(0.2147, device='cuda:0'), 'validation_loss_giou': tensor(0.6446, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.4838, device='cuda:0'), 'train_loss_ce': tensor(0.4119, device='cuda:0'), 'train_loss_bbox': tensor(0.1899, device='cuda:0'), 'train_loss_giou': tensor(0.5612, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.7380, device='cuda:0'), 'validation_loss_ce': tensor(0.4168, device='cuda:0'), 'validation_loss_bbox': tensor(0.2185, device='cuda:0'), 'validation_loss_giou': tensor(0.6143, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.4045, device='cuda:0'), 'train_loss_ce': tensor(0.4238, device='cuda:0'), 'train_loss_bbox': tensor(0.1777, device='cuda:0'), 'train_loss_giou': tensor(0.5461, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.7113, device='cuda:0'), 'validation_loss_ce': tensor(0.4276, device='cuda:0'), 'validation_loss_bbox': tensor(0.2102, device='cuda:0'), 'validation_loss_giou': tensor(0.6164, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(3.2236, device='cuda:0'), 'train_loss_ce': tensor(0.3068, device='cuda:0'), 'train_loss_bbox': tensor(0.2612, device='cuda:0'), 'train_loss_giou': tensor(0.8055, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.7548, device='cuda:0'), 'validation_loss_ce': tensor(0.4379, device='cuda:0'), 'validation_loss_bbox': tensor(0.2177, device='cuda:0'), 'validation_loss_giou': tensor(0.6142, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(3.0097, device='cuda:0'), 'train_loss_ce': tensor(0.4255, device='cuda:0'), 'train_loss_bbox': tensor(0.2340, device='cuda:0'), 'train_loss_giou': tensor(0.7071, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.7268, device='cuda:0'), 'validation_loss_ce': tensor(0.4118, device='cuda:0'), 'validation_loss_bbox': tensor(0.2167, device='cuda:0'), 'validation_loss_giou': tensor(0.6157, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.2934, device='cuda:0'), 'train_loss_ce': tensor(0.4336, device='cuda:0'), 'train_loss_bbox': tensor(0.1805, device='cuda:0'), 'train_loss_giou': tensor(0.4787, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.7026, device='cuda:0'), 'validation_loss_ce': tensor(0.4199, device='cuda:0'), 'validation_loss_bbox': tensor(0.2155, device='cuda:0'), 'validation_loss_giou': tensor(0.6026, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(2.2257, device='cuda:0'), 'train_loss_ce': tensor(0.4414, device='cuda:0'), 'train_loss_bbox': tensor(0.1801, device='cuda:0'), 'train_loss_giou': tensor(0.4420, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.8028, device='cuda:0'), 'validation_loss_ce': tensor(0.4215, device='cuda:0'), 'validation_loss_bbox': tensor(0.2218, device='cuda:0'), 'validation_loss_giou': tensor(0.6361, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
{'training_loss': tensor(3.2935, device='cuda:0'), 'train_loss_ce': tensor(0.4851, device='cuda:0'), 'train_loss_bbox': tensor(0.2494, device='cuda:0'), 'train_loss_giou': tensor(0.7807, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.7375, device='cuda:0'), 'validation_loss_ce': tensor(0.4192, device='cuda:0'), 'validation_loss_bbox': tensor(0.2163, device='cuda:0'), 'validation_loss_giou': tensor(0.6184, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')}
```

## Examples
{'size': tensor([512, 512]), 'image_id': tensor([1]), 'class_labels': tensor([4]), 'boxes': tensor([[0.2622, 0.5729, 0.0847, 0.0773]]), 'area': tensor([1717.9431]), 'iscrowd': tensor([0]), 'orig_size': tensor([1024, 1024])}

![Example](./example.png)