search

yiwenguo / Dynamic-Network-Surgery

Caffe implementation for dynamic network surgery.
Other
186 stars 70 forks source link

train result question #13

Closed renhui19931001 closed 7 years ago

renhui19931001 commented 7 years ago

using the dynamic network surgery ,I can train lenet5 on mnist database reach 99.07% accuracy,but when I
test the compression results,I find that Many parameters are not 0, but are very close to 0.The program did not write the corresponding zero program ? The training results from and downloaded caffemodel from your project there are some differences.can you tell me how to fix it or how to use the dynamic network surgery train lenet5 well?

renhui19931001 commented 7 years ago

OO,I have solve the problem:1.c_rate is the param related to ak,bk in paper,so we should not set 4 to the first convolution,and it cause the result of dosen't convergence.And about the weight ,I found that there is weightTmp in the compress_conv_layer,and it is used to do convolution but not weight.

chengsonghust commented 7 years ago

But how to get the weightTmp from caffemodel file?