MetaFormer is a general architecture abstracted from Transformers by not specifying the token mixer.
Source: MetaFormer Is Actually What You Need for VisionPaper | Code | Results | Date | Stars |
---|
Task | Papers | Share |
---|---|---|
Image Classification | 3 | 37.50% |
Object Detection | 2 | 25.00% |
Semantic Segmentation | 2 | 25.00% |
Fine-Grained Image Classification | 1 | 12.50% |
Component | Type |
|
---|---|---|
🤖 No Components Found | You can add them if they exist; e.g. Mask R-CNN uses RoIAlign |