search

zhaoweicai / hwgq

Caffe implementation of accurate low-precision neural networks
Other
118 stars 33 forks source link

How to save the binarized model? #6

Open iamweiweishi opened 5 years ago

iamweiweishi commented 5 years ago

According to my understanding, the weights of the trained model were not binarized directly. So, do I need to binarize the trained model with parameter 'scale' and then re-save the model?

zhaoweicai commented 5 years ago

Yes, and we have also provided the binarized models, please check README.

iamweiweishi commented 5 years ago

Hi, I trained one model using res50. And I tried to convert the trained models to binarized ones. Unfortunately, never succeed. I found the parameter 'alphas_' in 'binary_conv_layer.cpp', but I have no idea how to extract the parameter in trained models. Could you pls send me (nuptsww@126.com) some script example for reference. Thanks a lot.