Zhedong Zheng

@ronneycjm Yes, I didn't change. I think it is not the key for the effect.

@kevinrock change net.move('gpu') in 'test_vgg19.m' 16L to net.move('cpu')

@xiaojidan Is your GPU memory small? You may try using "net.conserveMemory = true" in test_vgg19.m Line13

@xiaojidan It's my fault. "net.conserveMemory = true" can not be used. (Because we still need information in certain net layer.) How big is your GPU memory?

@xiaojidan I use a 12GB TitanX. But I think 4GB is enough.

@xiaojidan My QQ is 397831949

Hi @rmxhhh Have you included the `$MATCONVNET/matlab` path? Under windows, you may need to change the file path from `/` to `\`.

@rmxhhh @xiaosuzhang Sorry. I did not make it clear. You need to make sure that you have included $This Repo/matlab$. I have included `+dagnn` under the `This Repo/matlab`.

Hi @feixiaoyu2223 Could you print ``` net.getParamIndex('fc751f') ``` It should be a positive number.

The code seems not find the weight `fc751f`, which is the weight for the last classification layer. Could you try ``` net = resnet52_market() net.getParamIndex('fc751f') ```