a mistake in the paper

[zw1992]

the output channel is mismatch with the output size ,maybe it should be 112.

[kuan-wang]

Thanks a lot! I've just updated the codes.
The gap of the model size is still large, welcome to report more bugs :)

[triangleCZH]

According to the paper right now, I believe they should be both 160 instead.
p.s. The paper has fixed the stride error in V3-small

[ujsyehao]

@kuan-wang
Hi, what are your mobilenet v3 small 1.0 params?
I calculate it and get 3.09 million (not include BN params), the paper claims 2.9 Million.

[kuan-wang]

The paper still has some bugs, now I renew the codes and the model size is close to paper.