| semantic-segmentation-on-ade20k | DeBiFormer-B (IN1k pretrain, Upernet 160k) | #95 | Validation mIoU: 52.0 |
| object-detection-on-coco-val2017 | DeBiFormer-B (Mask R-CNN 1x) | #255 | box AP: 48.5AP50: 70.2AP75: 53.3 |
| object-detection-on-coco-val2017 | DeBiFormer-S (Mask R-CNN 1x) | #272 | box AP: 47.5AP50: 69.7AP75: 52.1 |
| object-detection-on-coco-val2017 | DeBiFormer-B (RetinaNet 1x) | #279 | box AP: 47.1AP50: 68.2AP75: 50.2APS: 30.3APM: 51.1APL: 63.0 |
| object-detection-on-coco-val2017 | DeBiFormer-S (RetinaNet 1x) | #300 | box AP: 45.6AP50: 66.6AP75: 48.9APS: 28.7APM: 49.3APL: 61.6 |
| image-classification-on-imagenet | DeBiFormer-B | #348 | Top 1 Accuracy: 84.4% |
| image-classification-on-imagenet | DeBiFormer-S | #410 | Top 1 Accuracy: 83.9% |
| image-classification-on-imagenet | DeBiFormer-T | #636 | Top 1 Accuracy: 81.9% |