Jingkai Zhou

You can find more details in this link: https://blog.csdn.net/sophia_xw/article/details/102561410

Thanks, I have fixed this bug

"The learning rate is set to 0.1 with batch size 256 and decays to 1e-5 following the cosine schedule. " This line in paper means that I set 0.1 for...

> So the experimental results in your paper are obtained by training with batch size 256 or 1024? And if my training log matches yours in terms of the loss...

Yes, the Cuda code only speeds up small kernels due to the limited shared memory. You can modify the Cuda code to support large kernels.

"mul" means element-wise multiplication between spatial and channel filters "add" means element-wise addition between spatial and channel filters

1.支持组卷积；2.本来不同batch就不共享权重，不然为啥叫动态滤波器

是的

DDF may not work on windows, use Linux please.

Please move the *.so files to the ddf folder instead of the ddf.build folder