how to make the output of vl_simplenn has the same Matrix dimensions about caffe-vdsr HOT 4 CLOSED

Obviously the problem occurred in your trained network. Check the dimensions of output of each layer in res2 and you will know the problem.

from caffe-vdsr.

I just use the VDSR official code and trained model. I find the VDSR.mat can not be opened by Matlab, error info shown as below, but the VDSR_Official.mat in your code folder can be opened.

Caused by:
The CUDA driver was found, but the function 'cuDriverGetVersion' returned an error. The error returned was:
CUDA_ERROR_NOT_INITIALIZED

do you why?

from caffe-vdsr.

The parameters in VDSR.mat are stored in gpuArray format. If you don't have a gpu or there are some problem in the gpu usage, this problem will arise.

发自我的 iPhone

在 2016年9月25日，12:29，Peng Liu <[email protected]mailto:[email protected]> 写道：

I just use the VDSR official code and trained model. I find the VDSR.mat can not be opened by Matlab, error info shown as below, but the VDSR_Official.mat in your code folder can be opened.

Caused by:
The CUDA driver was found, but the function 'cuDriverGetVersion' returned an error. The error returned was:
CUDA_ERROR_NOT_INITIALIZED

do you why?

―
You are receiving this because you commented.
Reply to this email directly, view it on GitHubhttps://github.com//issues/12#issuecomment-249402136, or mute the threadhttps://github.com/notifications/unsubscribe-auth/ANB22oCpw_0F9SzaM5UpnUyLYExpcBXEks5qtfingaJpZM4KF0Bn.

from caffe-vdsr.

I got. Thank you ~!

from caffe-vdsr.