瓶颈变压器-火炬 在Pytorch中,在性能-计算权衡方面优于EfficientNet和DeiT的卷积(SotA)视觉识别模型的卷积+注意实现 安装 $ pip install bottleneck-transformer-pytorch 用法 import torch from torch import nn from bottleneck_transformer_pytorch import BottleStack layer = BottleStack ( dim = 256 , # channels in fmap_size = 64 , # feature map size dim_out = 2048 , # channels out proj_factor = 4 , #