Open XinDongol opened 6 years ago
Also interested in XNOR for ResNet! @XinDongol have you implemented it yourself?
Has anyone implemented XNOR for Resnet ?
I do not mind implementing ResNet for this if there is interest. However, I will not be able to train it. I think this task mostly just involves changing the class AlexNet here.
I have implemented XNOR-net for ResNet. Basically took the resnet.py
from torchvision and added the binary Convolution classes from this repo's alexnet.py
and then changed the convolutions in resnet.py
from nn.Conv2D to BinConv2D.
I have implemented ResNet-XNOR and reproduce result of xnor paper. I will release it soon when I have time.
@XinDongol could you kindly share the python models directory code for the ResNet-xNOR with me? Thanks!
I have implemented ResNet-XNOR and reproduce result of xnor paper. I will release it soon when I have time.
Could you please share it with me too? Thanks!
@XinDongol Would you like to share the python models directory code for the ResNet-XNOR or share the implementation idea?
@XinDongol Any update on the thread?
I have implemented the xnornet ++ paper with resnet-18. You can disable alpha, beta, gamma(scale factors for activation) in model file(BinConv2d). Then you can have the xnornet for resnet-18. Btw xnornet++ is improvement over xnornet and my implementation reaches 60% accuracy for 1 bit activation and 1 bit width. The link is https://github.com/legolas123/cv-tricks.com/tree/master/xnornet_plusplus
I was wondering whether there is XNOR for ResNet. Or can you give some hints about how to implement it?