Implementation of model compression with three knowledge distilling or teacher student methods .
The basic architecture is teacher-student model.
I used cifar-10 dataset to do this work.
Download cifar-10 dataset
In this the work, I use network in network as teacher model, lenet as student model.
The teacher model is pre-trained by caffe. And extract the model weight by .
Both network-in-network and lenet have little different from original model.
In docs, there are two images for the network architecture.
"teacher.npy" is the pre-trained model weights of teacher model.
"student.npy" is the model weights train on lenet, using ground turth label directly.
In teacher-student.py, there is three methods to train student network.
You need to modify the cifar-dataset-path in function read_cifar10
###Basic Usage train by 
python teacher-student.py --task train --model savemodel
train by 
python teacher-student.py --task train --model savemodel --noisy [--noisy_ratio --noisy_sigma]
train by 
python teacher-student.py --task train --model savemodel --KD [--lamda --tau]
**testing** >python teacher-student.py --task test --model trained_model
**validation** Also, you can validate your pre-trained teacher model by
> python teacher-student.py --task val
This can make sure that your caffe-teacher-model transfer to tensorflow successfully.
python teacher-student.py -h for more information
All three methods train 100 epochs, with dropout ratio=0.8, lr=1e-3, decay 0.1 at 80th epoch.
In method, noisy_ratio=0.5, sigma=0.1.
In methos, lamda=0.3, tau=0.3.
This table shows the accuracy on testing dataset, test by 100-epoch-model.
See more details in result.
The accuarcy of original model which directly learn by ground truth label:
teacher model : 78.1%
student model : 66.15%
 Ba, J. and Caruana, R. Do deep nets really need to be deep? In NIPS 2014.
 Bharat Bhusan Sau Vineeth N. Balasubramanian, Deep Model Compression: Distilling Knowledge from Noisy Teachers. arXiv 2016.
 Hinton, G. E., Vinyals, O., and Dean, J. Distilling the knowledge in a neural network. arXiv 2015.
 Network in Network model - https://github.com/aymericdamien/TensorFlow-Examples/
 Y. LeCun, L. Bottou, Y. Bengio and P. Haffner: Gradient-Based Learning Applied to Document Recognition, Proceedings of the IEEE 1998