backbone pretrain dataset paper download CaiT ImageNet 1k Going deeper with Image Transformers weights CoaT ImageNet 1k Co-Scale Conv-Attentional Image Transformers weights ConViT ImageNet 1k ConViT: Improving Vision Transformers with Soft Convolutional Inductive Biases weights CSPNet ImageNet 1k CSPNET: A NEW BACKBONE THAT CAN ENHANCE LEARNING CAPABILITY OF CNN cspresnet50 DenseNet ImageNet 1k weights dla60_res2net ImageNet 1k DLA Res2Net dla60_res2net DPN ImageNet 1k Dual Path Networks weights EfficientNet ImageNet 1k/21k EfficientNet-V2 tf_efficientnetv2_s_in21k GhostNet ImageNet 1k GhostNet ghostnet_100 CVPR 2020 gluon_resnet ImageNet 1k weights HRNet ImageNet 1k Deep High-Resolution Representation Learning for Visual Recognition weights TPAMI MLP-Mixer ImageNet 1k/21k MLP-Mixer: An all-MLP Architecture for Vision weights MobileNet V3 ImageNet 1k/21k Searching for MobileNetV3 weights ICCV 2019 NFNet, NF-RegNet, NF-ResNet ImageNet 1k Characterizing signal propagation to close the performance gap in unnormalized ResNets weights ICLR 2021 PiT ImageNet 1k Rethinking Spatial Dimensions of Vision Transformers weights RegNet ImageNet 1k Designing Network Design Spaces weights CVPR 2020 Res2Net,Res2NeXt ImageNet 1k Res2Net weights IEEE TPAMI 2021 ResNeSt ImageNet 1k ResNeSt: Split-Attention Networks weights ResNet ResNet v2 ImageNet 1k/21k weights ReXNet Rethinking Channel Dimensions for Efficient Model Design weights CVPR 2021 Selective Kernel Networks (ResNet base) ImageNet 1k Compounding the Performance Improvements of Assembled Techniques in a Convolutional Neural Network weights Swin Transformer ImageNet 1k/21k Swin Transformer: Hierarchical Vision Transformer using Shifted Windows weights Transformer in Transformer ImageNet 1k Transformer in Transformer weights TResNet ImageNet 1k/21k TResNet: High Performance GPU-Dedicated Architecture weights Twins ImageNet 1k Twins: Revisiting the Design of Spatial Attention in Vision Transformers weights ViT ImageNet 1k/21k Vit-Hybrid ImageNet 1k/21k weights