The problem about importance factor
developerFanYu opened this issue · comments
Hello. I am master in Xi'an Jiaotong Universtiy. This is an outstanding paper, but I have one question about importance factor.
I read the source code and find that the importance factor is defined as the weight sum of each block.
loss_bn_l1 = prune.cal_bn_l1_loss(get_prune_weights(model), FLAGS._bn_to_prune.penalty, rho)
Do you think my understanding is correct?
I can not understand that importance factor is the weight sum of each block.
Thank you very much!