There are 4 people in our group: Jiajun Duan, Hanwen Zhang, Ziming Zhang and Zewei Sheng. This code is made for our group, the original code is here.
Please make sure there is a folder named saved_model in the same path of the program.
The only thing that needs to do is to run the trauin_Le.py. It will begin at train the original LeNet_300_100 model and then pruning the model and pruning with compensating. Finally will test each model's accuracy.