Closed vzvzx closed 9 years ago
Hi, I try to train the net using this code , but I can not see the loss reduce. So I think if there is some init method?
I used the initial strategy described in http://arxiv.org/abs/1502.01852. It has been implemented in my caffe repository. You can just use it.
Hi, I try to train the net using this code , but I can not see the loss reduce. So I think if there is some init method?