📚 The doc issue

Does Swin impl in this repo support arbitrary dynamic execution-time-defined input resolution (same as other backbones)?

Initially Swin was trained to support only one resolution, but then hacks can be done to support arbitrary resolution. Two repos with such hacks:

• https://github.com/SwinTransformer/Swin-Transformer-Object-Detection/blob/master/mmdet/models/backbones/swin_transformer.py
• https://github.com/megvii-research/SOLQ/blob/main/models/swin_transformer.py

Related issues: microsoft/SimMIM#13 microsoft/esvit#17

cc @YosuaMichael