AojunZhou

how about pre-trained model performance and the 50% quantization performance?

@TwistedfateKing thanks, I don't do any experiments about bias quantization, I think you can convert bias from float-point into 8-bits, or remove bias directly.

hi, Q1: you can follow the paper "deep compression" or https://github.com/songhan/Deep-Compression-AlexNet. Q2: if you load image directly, the caffe can shuffle data, or you can use https://github.com/lim0606/caffe-googlenet-bn

@vassiliad thanks for your comments and suggestion. For first step, what about the max iteration you use, and the max iteration should make sure the learning rate decay to e^-5,...

@vassiliad you should fineturn more iterations and till the learning rate decay 0.00001, and you can speed up your training process with step_size = 2000 (~=1/3 epoch).

the 5 bits quantization? the sparse model partition must be careful, such as 0.05, 0.1,0.15,0.2,0.3,0.4,0.5,0.6,0.7,0.8,0.9,0.95,1.0. and more training epochs is useful.

yeah, I also just do INQ after DNS, only fixed the zero and using INQ to none-zero part.

No, we don't quantize BN layers to low-precision, In my code, you must skip the bn layers.

@KangolHsu you can skip the bn layers (include batch norm and scale layer), you can modify line 764 in "src/caffe/net.cpp" , we have added some comments in "src/caffe/net.cpp"

@zlheos have you completed all quantization step ?