🔼 PyramidTNT: Improved Transformer-in-Transformer Baselines with Pyramid Architecture PyramidTNT achieves better performances than the previous state-of-the-art vision transformers such as Swin Transformer Github: https://github.com/huawei-noah/CV-backbones Paper: https://arxiv.org/abs/2201.00978v1 GhostNet: github.com/huawei-…_pytorch @ai_machinelearning_big_data