How to do I train network for CHaiDNN

[guru3110]

I have trained the Alexnet Chaidnn model by removing the precision parameter. But when I am testing with the data set it always predicts same probability for all the images. But the same alexnet provided by caffe framework which is running on CPU is working well. I have compared CHaiDNN Alexnet model and Caffe Alexnet model in the netscope tool. I found that CHaiDNN model doesn't support LRN type. norm1, norm2 and drop6 and drop7 are the extra parameters in the caffe model. These parameter play the major role in training phase. How should i replicate these layer in the CHaiDNN model. Kindly help to fix this issue.