Is DepthWiseConv2dImplicitGEMM absolutely necessary?

DingXiaoH / RepLKNet-pytorch

Scaling Up Your Kernels to 31x31: Revisiting Large Kernel Design in CNNs (CVPR 2022)

MIT License

867 stars 86 forks source link

Closed Harvey-Mei closed 2 years ago

Harvey-Mei commented 2 years ago

Hello DingXiao Can we just use the Conv2D in torch? Are there other effects besides computational efficiency?

DingXiaoH commented 2 years ago

Of course you can use nn.Conv2d. Our implementation is just faster.

Harvey-Mei commented 2 years ago

OK. Thanks for your reply!