caiwenpu

It's same to me. I could only get ~91% accuracy for the MobileNetV2 in this code. I think this is because the downsample layers are too many(three downsample layers), leading...

Hi, sorry to bother you again. I found the grad of full-precision weights is set by the following code ``` grad_fp_weight = pos * grad_ternary_weight * pos_indices + \ grad_ternary_weight...

Hi, thanks for your careful reply. Maybe my expression is not very clear. More carefully, the gradient of negative pre-ternary weights are set to `neg * grad_ternary_weight * neg_indices` in...

I use Theano with version 0.9.0 and I have sloved the problem by add an argument " impl='c' " in function make_thunk of class Gemm and class XnorGemm.