Fetching the paper…

Conv2Former: A Simple Transformer-Style ConvNet for Visual Recognition · Around