can you explain how your method works?

[gengwb]

Hi, lusenkong
I am very glad that you share your code with us, can you explain how u algorithm works？
can u share some demo about how to use Caffemodel_Compress?

[MedicineStick]

I used VS2013 as my local IDE,beyond my Company's limitation ,I'm afraid I couldn't share my datum with you. here is some tips .

1, Caffemodel_Compress is based on compress theory, which is cleaning less-significant convolutional filters up with your caffemodel, this technique calculates each filter weights' L1 or L2 ,then sorting and wiping away less-significant one
2, Two files that you need before getting start, 1) your caffemodel 2) your prototxt(test) , the XML file as you see in repsitory is the params during progress running, including the Convolutin layer which you need to pruning and how much convolutional filters that you may want to wipe away.
(formula : num_output * cut_ratio = cut_num )

if there are any questions during your deployment , come and pull requests to me , you are always be welcomed !