Fetching the paper…

UniFormer: Unifying Convolution and Self-attention for Visual Recognition · Around