Fetching the paper…

ConvNets vs. Transformers: Whose Visual Representations are More Transferable? · Around