CondConv-tensorflow

Conditional convolution (Dynamic convolution) in tensorflow2.2.0. This depository implements the method described in the paper:

CondConv: Conditionally Parameterized Convolutions for Efficient Inference
Brandon Yang, Gabriel Bender, Quoc V.Le, Jiquan Ngiam
Source PDF

Meanwhile, the softmax with a large temperature for kernel attention introduced by Dynamic Convolution: Attention Over Convolution Kernels is adopted.

Another similar paper: DyNet: Dynamic Convolution for Accelerating Convolutional Neural Networks.

Start

You can start according to the default arguments by python main.py. Or specify the arguments:

python main.py --arch cond_cifar_resnet --num_layers 56 --num_experts 3 --dataset cifar10 --num_classes 10