| image-classification-on-imagenet | gSwin-S | – | Top 1 Accuracy: 83.01%Number of params: 39.8MGFLOPs: 7.0 |
| image-classification-on-imagenet | gSwin-T | – | Top 1 Accuracy: 81.71%Number of params: 21.8MGFLOPs: 3.6 |
| image-classification-on-imagenet | gSwin-VT | – | Top 1 Accuracy: 80.32%Number of params: 15.5MGFLOPs: 2.3 |
| instance-segmentation-on-coco-test-dev | gSwin-S | – | mask AP: 45.03 |
| instance-segmentation-on-coco-test-dev | gSwin-T | – | mask AP: 44.16 |
| instance-segmentation-on-coco-test-dev | gSwin-VT | – | mask AP: 42.87 |
| semantic-segmentation-on-ade20k-val | gSwin-S | – | mIoU: 49.69Pixel Accuracy: 83.43 |
| semantic-segmentation-on-ade20k-val | gSwin-T | – | mIoU: 47.63Pixel Accuracy: 82.60 |
| semantic-segmentation-on-ade20k-val | gSwin-VT | – | mIoU: 45.07Pixel Accuracy: 81.79 |